**Help a leading AI lab define what great design looks like, and build the prototypes that prove it.**

# 1\. Overview

A leading AI lab is looking for senior product designers and design-minded engineers with exceptional taste to help shape how its AI models design and build for the web and for games. You'll work directly with the lab's researchers to define what "great" looks like in UI and UX, and you'll build high-quality prototypes that show it. If you care deeply about great interfaces, like to prototype in code or with AI tools, and can explain why one design works better than another, this role is for you. This is a full-time commitment of 40 hours per week.

This is a W-2 employment position with Cincinnatus LLC, with the opportunity to be placed at a leading AI lab as part of their extended workforce.

# 2\. Key Responsibilities

- Partner with AI researchers to identify best-in-class UI and UX patterns, and turn them into clear principles and rubrics that define what great design looks like.
    
- Build polished, interactive web prototypes, in code or with AI-native tools such as Cursor, v0, Lovable or Bolt, that showcase strong and diverse design patterns across many product categories.
    
- Review AI-generated interfaces and prototypes, and give specific, actionable written feedback on layout, hierarchy, interaction, motion and visual polish.
    
- Explore a wide range of visual styles, frameworks and animation approaches, from focused productivity tools to playful consumer apps and games.
    
- Collaborate with other designers and researchers to keep standards consistent and keep raising the quality bar.
    

# 3\. Core Qualifications

- 7+ years of experience as a product designer, design lead, design engineer, or creative front-end engineer at a product- or design-led company (for example Apple, Airbnb, Stripe, Figma, Linear, Spotify or Notion) or a top design agency.
    
- Hands-on prototyping skills: you build working web prototypes yourself, in code (HTML, CSS, JavaScript, React, Tailwind) or with AI-native tools (Cursor, v0, Lovable, Bolt, Framer).
    
- A strong grounding in user-centered design, interaction design and rapid prototyping.
    
- Senior-level design judgment, and the confidence to set direction and explain clearly why a design works.
    
- Ability to engage reliably for 40 hours/week during weekdays.
    
- Clear written communication and the ability to give precise, constructive design feedback.
    

**Nice to have:** Fluency with motion and animation (Framer Motion, GSAP, Rive, Lottie, After Effects) or 3D and WebGL, and game design experience in 2D or 3D. You don't need all of these to apply.

**About Cincinnatus LLC:**
Cincinnatus LLC is an enterprise staffing company that partners with leading technology companies to source and employ highly skilled professionals for contingent and contract-based opportunities. Cincinnatus serves as the employer of record for these engagements, providing W-2 employment, payroll, benefits, and compliance, while placing employees directly within client teams to work on high-impact initiatives.

**Equal Employment Opportunity:**
Cincinnatus is proud to be an Equal Employment Opportunity employer. We do not discriminate based upon race, religion, color, national origin, sex (including pregnancy, childbirth, reproductive health decisions, or related medical conditions), sexual orientation, gender identity, gender expression, age, status as a protected veteran, status as an individual with a disability, genetic information, political views or activity, or any other legally protected characteristic.

Senior Product Designer, UI/UX

## About the work

**CritPt** is a public benchmark of research-level physics challenges, built to test whether frontier AI models can carry out genuine physics research reasoning rather than textbook problem solving. The benchmark paper is **[arXiv:2509.26574](https://arxiv.org/abs/2509.26574)** and we recommend reading it before applying. It will tell you quickly whether this work interests you.

We are engaging physicists to work on research-level physics problems in their own subfield. Depending on where your publication record fits, that can mean creating problems, solving them, reviewing completed work, or auditing it. We agree the specific assignment with you once you are matched to an area.

This is research-grade work rather than volume work. Whatever you produce has to be complete enough for another specialist in your subfield to follow and verify independently, so written reasoning is part of every assignment.

## Research areas in this panel

One area. We match narrowly: you need to have published on this specific phenomenon, not in biophysics broadly.

**1. Stochastic autocatalytic growth: chemical master equation, branching processes:** Autocatalytic reaction cycles, stochastic Hinshelwood cycle, chemical master equation processes, circulant matrix spectra, transient oscillations toward balanced exponential growth, moment hierarchies for reaction networks, generating-function methods.

If you also work on long-range dispersal, heavy-tailed jump processes or front propagation in range expansions, say so on the form. A closely related area sits in our statistical physics panel and draws on the same methods.

## Methods we expect to find in your own publications

You should be able to point to your own papers demonstrating **at least one** of the following families:

- Chemical master equation and the stochastic Hinshelwood cycle
- Multi-type branching processes and circulant matrix spectra
- Moment hierarchies for reaction networks
- Generating-function methods

## Who we are looking for

**A PhD in biophysics, statistical physics, applied mathematics or a closely related field.** This is a hard requirement. Postdoctoral researchers, research scientists and junior faculty are the strongest fit. Senior PhD students with a strong first-author record are welcome to apply.

**Published work on stochastic growth or reaction-network dynamics specifically.** This is the single most common reason we decline otherwise excellent researchers. Command of the methods is not enough if you have not published on the phenomenon itself.

**A verifiable publication record.** Three to five representative papers with arXiv IDs or DOIs, ideally from the last five years. First author strongly preferred. Every paper you list will be checked against the public record.

**Working proficiency with LaTeX, Python, SymPy and Jupyter.** Some familiarity with an agentic coding extension in VS Code is useful. Gaps there are acceptable if you declare them honestly.

**English at B2 or above**, including written reasoning. A large part of the value you add is how clearly you set out your argument.

## Application steps

1. **Apply and complete the attached form.** Basic information, education, research experience, your method self-attestation, and the area above if you are a fit. Give an arXiv ID or DOI of **your own paper** as proof, with your author position and the methods it demonstrates. A selection without proof is not scored.
2. **We verify your papers and authorship** against the public record.
3. **Then one of two things happens.** Either we onboard you directly, or we invite you to a short live alignment call to agree the area and the assignment with you.
4. A brief **30 to 45 minute assessment** may be added, but only where we need it. Most applicants will not see one.

## Commitment and rate

**10 hours per week**, sustained across an **8 to 10 week** window, starting immediately. Remote and asynchronous with no fixed hours.

**$80 to $110 per hour**, set by depth of subdomain match.

Stochastic Biophysics Physicist (PhD)

## About the work

**CritPt** is a public benchmark of research-level physics challenges, built to test whether frontier AI models can carry out genuine physics research reasoning rather than textbook problem solving. The benchmark paper is **[arXiv:2509.26574](https://arxiv.org/abs/2509.26574)** and we recommend reading it before applying. It will tell you quickly whether this work interests you.

We are engaging physicists to work on research-level physics problems in their own subfield. Depending on where your publication record fits, that can mean creating problems, solving them, reviewing completed work, or auditing it. We agree the specific assignment with you once you are matched to an area.

This is research-grade work rather than volume work. Whatever you produce has to be complete enough for another specialist in your subfield to follow and verify independently, so written reasoning is part of every assignment.

## Research areas in this panel

Two areas, both on the linear stability of buoyancy-driven convection. We match narrowly: you need to have published on convection stability specifically, not on fluid dynamics broadly. Each area lists the methods it requires.

**1. Rayleigh-Benard convection: linear stability with mixed boundary conditions:** Rayleigh-Benard convection, Boussinesq approximation, linear stability analysis, normal-mode perturbation expansion, mixed mechanical and thermal boundary conditions, fixed-flux thermal forcing, marginal stability analysis, spectral eigenvalue methods.

**2. Porous-medium (Rayleigh-Darcy) convection: linear stability:** Darcy's law for porous-medium flow, Rayleigh-Darcy convection, Boussinesq buoyancy approximation, linear stability analysis, constant-flux and open boundary conditions, normal-mode perturbation expansion, marginal stability analysis, buoyancy-driven transport in saturated porous layers.

## Methods we expect to find in your own publications

You should be able to point to your own papers demonstrating **at least one** of the following families:

- Boussinesq approximation and buoyancy-driven flow
- Linear stability analysis and normal-mode perturbation expansion
- Marginal stability analysis and critical Rayleigh number determination
- Spectral eigenvalue methods
- Darcy's law and porous-medium flow

## Who we are looking for

**A PhD in fluid mechanics, applied mathematics, mechanical engineering or a closely related field.** This is a hard requirement. Postdoctoral researchers, research scientists and junior faculty are the strongest fit. Senior PhD students with a strong first-author record are welcome to apply.

**Published work on convection stability specifically.** This is the single most common reason we decline otherwise excellent researchers. Command of the methods is not enough if you have not published on the phenomenon itself.

**A verifiable publication record.** Three to five representative papers with DOIs or arXiv IDs, ideally from the last five years. First author strongly preferred. A journal DOI is expected here, since much of this literature is not on arXiv. Every paper you list will be checked against the public record.

**Working proficiency with LaTeX, Python and Jupyter.** Some familiarity with an agentic coding extension in VS Code is useful. Gaps there are acceptable if you declare them honestly.

**English at B2 or above**, including written reasoning. A large part of the value you add is how clearly you set out your argument.

## Application steps

1. **Apply and complete the attached form.** Basic information, education, research experience, your method self-attestation, and the areas above that you are the best fit for. For each area you select, give a DOI or arXiv ID of **your own paper** as proof, with your author position and the methods it demonstrates. A selection without proof is not scored.
2. **We verify your papers and authorship** against the public record.
3. **Then one of two things happens.** Either we onboard you directly, or we invite you to a short live alignment call to agree the area and the assignment with you.
4. A brief **30 to 45 minute assessment** may be added, but only where we need it. Most applicants will not see one.

## Commitment and rate

**10 hours per week**, sustained across an **8 to 10 week** window, starting immediately. Remote and asynchronous with no fixed hours.

**$80 to $110 per hour**, set by depth of subdomain match.

Fluid Dynamics and Convection Stability Physicist (PhD)

## About the work

**CritPt** is a public benchmark of research-level physics challenges, built to test whether frontier AI models can carry out genuine physics research reasoning rather than textbook problem solving. The benchmark paper is **[arXiv:2509.26574](https://arxiv.org/abs/2509.26574)** and we recommend reading it before applying. It will tell you quickly whether this work interests you.

We are engaging physicists to work on research-level physics problems in their own subfield. Depending on where your publication record fits, that can mean creating problems, solving them, reviewing completed work, or auditing it. We agree the specific assignment with you once you are matched to an area.

This is research-grade work rather than volume work. Whatever you produce has to be complete enough for another specialist in your subfield to follow and verify independently, so written reasoning is part of every assignment. In this panel the standard of proof sits closer to mathematics than to physics.

## Research areas in this panel

Four areas. We match narrowly: you need to have published on one of these specific phenomena, not in mathematical physics broadly. Each area lists the methods it requires.

**1. Special functions: hypergeometric identities and analytic continuation:** Gaussian hypergeometric functions, parameter differentiation of special functions, quadratic transformation identities for hypergeometric series, analytic continuation in a series index, digamma-function series, asymptotic and series expansions of special functions.

**2. Integrable systems: half-wave maps, Lax pairs, Haldane-Shastry and Calogero-Moser:** Half-wave maps equation, Lax pair formalism and isospectral flows, Hilbert transform and singular integral operators, classical Haldane-Shastry spin chain, Calogero-Moser systems, hierarchies of conserved charges, solitons of integrable spin-field equations, numerical quadrature of singular integrals.

**3. Permutation combinatorics for random tensor network entanglement (Cayley distance, Weingarten):** Cayley distance and geodesics in the symmetric group, minimal factorizations of permutations, non-crossing partitions, replica permutation spin models, random tensor network entanglement entropy, Weingarten calculus, multipartite entanglement measures, combinatorial enumeration of configurations.

**4. Conformal geometry: Fefferman-Graham ambient metric, obstruction tensors:** Fefferman-Graham ambient metric construction, conformal geometry and Weyl covariance, extended obstruction tensors, Schouten and Weyl curvature tensors, order-by-order solution of Ricci-flatness conditions, poles and residues under dimensional continuation, Poincare-Einstein asymptotic expansions.

## Methods we expect to find in your own publications

You should be able to point to your own papers demonstrating **at least one** of the following families:

- Special function analysis: Gaussian hypergeometric functions, parameter differentiation, quadratic transformation identities, analytic continuation in a series index, asymptotic expansions
- Integrability: Lax pair formalism, isospectral flows, Hilbert transform and singular integral operators, hierarchies of conserved charges, numerical quadrature of singular integrals
- Combinatorial: Cayley distance and geodesics in the symmetric group, minimal factorizations of permutations, non-crossing partitions, Weingarten calculus, multipartite entanglement measures
- Geometric: Fefferman-Graham ambient metric construction, Weyl covariance, Schouten and Weyl curvature tensors, order-by-order solution of Ricci-flatness conditions, Poincare-Einstein asymptotic expansions

## Who we are looking for

**A PhD in mathematical physics, theoretical physics or mathematics.** This is a hard requirement. Postdoctoral researchers, research scientists and junior faculty are the strongest fit. Senior PhD students with a strong first-author record are welcome to apply.

**Published work on the specific phenomenon above, not the adjacent one.** This is the single most common reason we decline otherwise excellent physicists. Command of the methods is not enough if you have not published on the phenomenon itself.

**A verifiable publication record.** Three to five representative papers with arXiv IDs or DOIs, ideally from the last five years. First author strongly preferred. Every paper you list will be checked against the public record.

**Working proficiency with LaTeX, Python, SymPy and Jupyter.** Symbolic computation matters more in this panel than in most. Some familiarity with an agentic coding extension in VS Code is useful. Gaps there are acceptable if you declare them honestly.

**English at B2 or above**, including written reasoning. A large part of the value you add is how clearly you set out your argument.

## Application steps

1. **Apply and complete the attached form.** Basic information, education, research experience, your method self-attestation, and up to five of the areas above that you are the best fit for. For each area you select, give an arXiv ID or DOI of **your own paper** as proof, with your author position and the methods it demonstrates. A selection without proof is not scored.
2. **We verify your papers and authorship** against the public record.
3. **Then one of two things happens.** Either we onboard you directly, or we invite you to a short live alignment call to agree the area and the assignment with you.
4. A brief **30 to 45 minute assessment** may be added, but only where we need it. Most applicants will not see one.

## Commitment and rate

**10 hours per week**, sustained across an **8 to 10 week** window, starting immediately. Remote and asynchronous with no fixed hours.

**$80 to $110 per hour**, set by depth of subdomain match.

Mathematical Physicist (PhD)

## About the work

**CritPt** is a public benchmark of research-level physics challenges, built to test whether frontier AI models can carry out genuine physics research reasoning rather than textbook problem solving. The benchmark paper is **[arXiv:2509.26574](https://arxiv.org/abs/2509.26574)** and we recommend reading it before applying. It will tell you quickly whether this work interests you.

We are engaging physicists to work on research-level physics problems in their own subfield. Depending on where your publication record fits, that can mean creating problems, solving them, reviewing completed work, or auditing it. We agree the specific assignment with you once you are matched to an area.

This is research-grade work rather than volume work. Whatever you produce has to be complete enough for another specialist in your subfield to follow and verify independently, so written reasoning is part of every assignment.

## Research areas in this panel

Five areas. We match narrowly: you need to have published on one of these specific phenomena, not in statistical physics broadly. Each area lists the methods it requires.

**1. Disordered spin models and the Nishimori line, QEC thresholds, mixed-state topological order:** Replicated random-bond Ising model, Nishimori line and gauge symmetry of disordered spin models, quenched disorder averaging, Kramers-Wannier duality and square-lattice self-duality, Ashkin-Teller coupled Ising flavors, domain-wall free energy from twisted boundary conditions, exact lattice evaluation of Ising partition functions, decoherence-induced mixed-state topological order and quantum memory thresholds.

**2. Entanglement entropy via the replica trick, Goldstone modes and tower of states:** Replica trick, analytic continuation in the replica index, Gaussian lattice sums and theta functions, Poisson summation and modular transformations, Renyi and von Neumann entanglement entropies, Goldstone modes and tower-of-states structure.

**3. Range expansions with long-range dispersal, front propagation, heavy-tailed jumps:** Long-range dispersal kernels, heavy-tailed jump processes, stratified diffusion and coalescing colony models, self-consistent mean-field growth equations, convolution integral equations, asymptotic expansions with logarithmic corrections, front propagation in range expansions.

**4. Lattice path enumeration and entanglement domain-wall statistical mechanics:** Lattice random walks with internal states, generating function methods, first-return decomposition of lattice paths, transfer-matrix partition functions, splitting and recombination of composite excitations, entanglement domain-wall statistical mechanics, weighted path enumeration.

**5. Two-dimensional CFT: Ising minimal model, BPZ equations, conformal blocks, Coulomb gas:** Two-dimensional conformal field theory, Ising minimal model, Virasoro algebra and degenerate representations, BPZ null-vector differential equations, conformal blocks and crossing symmetry, operator product expansion and fusion rules, Coulomb gas integral representations, free-fermion and bosonization methods.

## Methods we expect to find in your own publications

You should be able to point to your own papers demonstrating **at least one** of the following families:

- Exact lattice methods: Kramers-Wannier duality, self-duality, transfer-matrix partition functions, exact evaluation of coupled Ising flavors
- Replica and field-theoretic: replica trick and analytic continuation in the replica index, Gaussian lattice sums and theta functions, Renyi and von Neumann entropies
- Asymptotic analysis: convolution integral equations, asymptotic expansions with logarithmic corrections, stratified diffusion, generating function methods
- Conformal: Virasoro algebra, degenerate representations, BPZ null-vector differential equations, operator product expansion and fusion rules, free-fermion and bosonization methods

## Who we are looking for

**A PhD in statistical physics, mathematical physics or a closely related field.** This is a hard requirement. Postdoctoral researchers, research scientists and junior faculty are the strongest fit. Senior PhD students with a strong first-author record are welcome to apply.

**Published work on the specific phenomenon above, not the adjacent one.** This is the single most common reason we decline otherwise excellent physicists. Command of the methods is not enough if you have not published on the phenomenon itself.

**A verifiable publication record.** Three to five representative papers with arXiv IDs or DOIs, ideally from the last five years. First author strongly preferred. Every paper you list will be checked against the public record.

**Working proficiency with LaTeX, Python, SymPy and Jupyter.** Some familiarity with an agentic coding extension in VS Code is useful. Gaps here are acceptable if you declare them honestly.

**English at B2 or above**, including written reasoning. A large part of the value you add is how clearly you set out your argument.

## Application steps

1. **Apply and complete the attached form.** Basic information, education, research experience, your method self-attestation, and up to five of the areas above that you are the best fit for. For each area you select, give an arXiv ID or DOI of **your own paper** as proof, with your author position and the methods it demonstrates. A selection without proof is not scored.
2. **We verify your papers and authorship** against the public record.
3. **Then one of two things happens.** Either we onboard you directly, or we invite you to a short live alignment call to agree the area and the assignment with you.
4. A brief **30 to 45 minute assessment** may be added, but only where we need it. Most applicants will not see one.

## Commitment and rate

**10 hours per week**, sustained across an **8 to 10 week** window, starting immediately. Remote and asynchronous with no fixed hours.

**$80 to $110 per hour**, set by depth of subdomain match.

Statistical Physics and Thermodynamics Physicist (PhD)

## About the work

**CritPt** is a public benchmark of research-level physics challenges, built to test whether frontier AI models can carry out genuine physics research reasoning rather than textbook problem solving. The benchmark paper is **[arXiv:2509.26574](https://arxiv.org/abs/2509.26574)** and we recommend reading it before applying. It will tell you quickly whether this work interests you.

We are engaging physicists to work on research-level physics problems in their own subfield. Depending on where your publication record fits, that can mean creating problems, solving them, reviewing completed work, or auditing it. We agree the specific assignment with you once you are matched to an area.

This is research-grade work rather than volume work. Whatever you produce has to be complete enough for another specialist in your subfield to follow and verify independently, so written reasoning is part of every assignment.

## Research areas in this panel

Four areas, three in torsionful gravity and axion inflation and one in high-energy astrophysics. We match narrowly: you need to have published on one of these specific phenomena, not in gravitation broadly. Each area lists the methods it requires.

**1. Einstein-Cartan and Nieh-Yan axion inflation: perturbations and power spectrum:** Einstein-Cartan-Palatini first-order gravity, Nieh-Yan topological invariant, spacetime torsion in the tetrad formalism, pseudoscalar axion inflation, natural inflation potentials, cosmological perturbation theory, Mukhanov-Sasaki mode equation, primordial curvature power spectrum, slow-roll approximation.

**2. Einstein-Cartan and Nieh-Yan axion inflation: FRW background dynamics:** First-order Palatini formulation of gravity, tetrad and spin-connection variables, Einstein-Cartan gravity with torsion, Nieh-Yan topological term, axionic natural inflation potential, FRW background cosmological dynamics, numerical integration of background field equations.

**3. Dynamical Chern-Simons gravity in Palatini formalism, parity-violating inflation:** First-order Palatini and Einstein-Cartan formulations of gravity, tetrad and spin-connection variables, spacetime torsion and contorsion decomposition, gravitational Chern-Simons coupling of a pseudoscalar to the Pontryagin density, parity violation in the gravitational sector, FRW background cosmology and slow-roll inflation, numerical integration of coupled cosmological field equations.

**4. High-energy astrophysics: blazar jets, photohadronic emission, multimessenger constraints:** Relativistic jet beaming and Doppler boosting, photopion production at the Delta resonance, photohadronic interaction optical depth, electromagnetic cascade emission, synchrotron target photon fields, one-zone blazar emission modeling, variability-based causality constraints, multimessenger X-ray constraints.

## Methods we expect to find in your own publications

You should be able to point to your own papers demonstrating **at least one** of the following families:

- First-order gravity: Einstein-Cartan-Palatini formalism, tetrad and spin-connection variables, spacetime torsion and contorsion decomposition
- Topological terms in gravity: Nieh-Yan topological invariant, gravitational Chern-Simons coupling to the Pontryagin density
- Cosmological perturbation theory: Mukhanov-Sasaki mode equation, primordial curvature power spectrum, slow-roll approximation
- Numerical cosmology: integration of coupled background field equations in modified gravity
- Astroparticle modelling: relativistic beaming and Doppler boosting, photohadronic processes, electromagnetic cascades, one-zone emission models

## Who we are looking for

**A PhD in gravitation, cosmology, astroparticle physics or a closely related field.** This is a hard requirement. Postdoctoral researchers, research scientists and junior faculty are the strongest fit. Senior PhD students with a strong first-author record are welcome to apply.

**Published work on the specific phenomenon above, not the adjacent one.** This is the single most common reason we decline otherwise excellent physicists. Working in modified gravity is not enough if your papers are not on torsion, Nieh-Yan or Chern-Simons terms. Command of the numerical methods is not enough if you have not published on the phenomenon itself.

**A verifiable publication record.** Three to five representative papers with arXiv IDs or DOIs, ideally from the last five years. First author strongly preferred. Every paper you list will be checked against the public record.

**Sign-convention discipline.** Work in this area turns on tetrad and spin-connection conventions, and we are looking for people who are precise about them.

**Working proficiency with LaTeX, Python, SymPy and Jupyter.** Some familiarity with an agentic coding extension in VS Code is useful. Gaps here are acceptable if you declare them honestly.

**English at B2 or above**, including written reasoning. A large part of the value you add is how clearly you set out your argument.

## Application steps

1. **Apply and complete the attached form.** Basic information, education, research experience, your method self-attestation, and up to five of the areas above that you are the best fit for. For each area you select, give an arXiv ID or DOI of **your own paper** as proof, with your author position and the methods it demonstrates. A selection without proof is not scored.
2. **We verify your papers and authorship** against the public record.
3. **Then one of two things happens.** Either we onboard you directly, or we invite you to a short live alignment call to agree the area and the assignment with you.
4. A brief **30 to 45 minute assessment** may be added, but only where we need it. Most applicants will not see one.

## Commitment and rate

**10 hours per week**, sustained across an **8 to 10 week** window, starting immediately. Remote and asynchronous with no fixed hours.

**$80 to $110 per hour**, set by depth of subdomain match.

Gravitation, Cosmology and Astrophysics Physicist (PhD)

## About the work

**CritPt** is a public benchmark of research-level physics challenges, built to test whether frontier AI models can carry out genuine physics research reasoning rather than textbook problem solving. The benchmark paper is **[arXiv:2509.26574](https://arxiv.org/abs/2509.26574)** and we recommend reading it before applying. It will tell you quickly whether this work interests you.

We are engaging physicists to work on research-level physics problems in their own subfield. Depending on where your publication record fits, that can mean creating problems, solving them, reviewing completed work, or auditing it. We agree the specific assignment with you once you are matched to an area.

This is research-grade work rather than volume work. Whatever you produce has to be complete enough for another specialist in your subfield to follow and verify independently, so written reasoning is part of every assignment.

## Research areas in this panel

Seven areas. We match narrowly: you need to have published on one of these specific phenomena, not in AMO broadly. Each area lists the methods it requires.

**1. Strong-field and high-harmonic generation, structured light and angular momentum:** High-harmonic generation, orbital angular momentum of structured light, spin angular momentum and optical helicity, angular momentum conservation selection rules, self-torque of light, strong-field light-matter interaction.

**2. Levitated optomechanics: optical binding and dipole-dipole coupled oscillators:** Levitated optomechanics, optical tweezers, optical binding, light-induced dipole-dipole interactions, Rayleigh point-dipole approximation, Gaussian beam propagation, nonreciprocal coupling, coupled harmonic oscillator normal modes.

**3. Levitated optomechanics: torsional and librational modes of anisotropic nanoparticles:** Levitated optomechanics, torsional and librational modes, anisotropic polarizability of dielectric ellipsoids, optical tweezers, light-induced dipole-dipole torques, rigid-body rotational dynamics, harmonic quantization of small oscillations, beam-splitter coupling Hamiltonians.

**4. Precision measurement: Penning-trap quantum cyclotron, cavity QED radiative shifts:** Geonium theory of Penning traps, single-electron quantum cyclotron, cavity quantum electrodynamics mode structure, quantization of the radiation field, non-relativistic perturbation theory, dipole approximation, radiative frequency shifts and self-energy subtraction, electron magnetic moment tests of quantum electrodynamics.

**5. Ultracold atoms in optical lattices: tight-binding and Wannier parametrization:** Optical lattice potentials from laser interference, AC Stark shift and atomic polarizability, tight-binding lattice Hamiltonians, Wannier function formalism, harmonic approximation of lattice wells, s-wave contact pseudopotential, recoil energy and deep-lattice expansion, quantum simulation with ultracold fermions.

**6. Cavity QED: Jaynes-Cummings, dark states, open-system steady states:** Cavity quantum electrodynamics, Jaynes-Cummings interaction, bright and dark atomic states, Lindblad master equations, spontaneous emission into free space, coherent states of the radiation field, steady states of open quantum systems, photon-number coherences and decoherence.

**7. Few-body physics: Efimov effect, hyperspherical methods, zero-range universality:** Efimov effect, hyperspherical coordinates and hyperangular channel functions, zero-range Bethe-Peierls boundary conditions, discrete scale invariance in the three-body problem, bosonic permutation symmetrization, universality at large scattering length, wave-function overlap integrals.

## Methods we expect to find in your own publications

You should be able to point to your own papers demonstrating **at least one** of the following families:

- Light-matter: high-harmonic generation, angular momentum conservation selection rules, strong-field light-matter interaction, AC Stark shift and atomic polarizability
- Trap and oscillator dynamics: optical tweezers, Gaussian beam propagation, rigid-body rotational dynamics, harmonic quantization of small oscillations, beam-splitter coupling Hamiltonians
- Open quantum systems: Lindblad master equations, spontaneous emission into free space, photon-number coherences and decoherence
- Precision QED: non-relativistic perturbation theory, dipole approximation, self-energy subtraction, radiative frequency shifts
- Lattice and few-body: Wannier function formalism, s-wave contact pseudopotential, hyperspherical coordinates, zero-range boundary conditions

## Who we are looking for

**A PhD in atomic, molecular or optical physics or a closely related field.** This is a hard requirement. Postdoctoral researchers, research scientists and junior faculty are the strongest fit. Senior PhD students with a strong first-author record are welcome to apply.

**Published work on the specific phenomenon above, not the adjacent one.** This is the single most common reason we decline otherwise excellent physicists. Command of the methods is not enough if you have not published on the phenomenon itself.

**A verifiable publication record.** Three to five representative papers with arXiv IDs or DOIs, ideally from the last five years. First author strongly preferred. Every paper you list will be checked against the public record.

**Working proficiency with LaTeX, Python, SymPy and Jupyter.** Some familiarity with an agentic coding extension in VS Code is useful. Gaps here are acceptable if you declare them honestly.

**English at B2 or above**, including written reasoning. A large part of the value you add is how clearly you set out your argument.

## Application steps

1. **Apply and complete the attached form.** Basic information, education, research experience, your method self-attestation, and up to five of the areas above that you are the best fit for. For each area you select, give an arXiv ID or DOI of **your own paper** as proof, with your author position and the methods it demonstrates. A selection without proof is not scored.
2. **We verify your papers and authorship** against the public record.
3. **Then one of two things happens.** Either we onboard you directly, or we invite you to a short live alignment call to agree the area and the assignment with you.
4. A brief **30 to 45 minute assessment** may be added, but only where we need it. Most applicants will not see one.

## Commitment and rate

**10 hours per week**, sustained across an **8 to 10 week** window, starting immediately. Remote and asynchronous with no fixed hours.

**$80 to $110 per hour**, set by depth of subdomain match.

Atomic, Molecular and Optical Physicist (PhD)

## About the work

**CritPt** is a public benchmark of research-level physics challenges, built to test whether frontier AI models can carry out genuine physics research reasoning rather than textbook problem solving. The benchmark paper is **[arXiv:2509.26574](https://arxiv.org/abs/2509.26574)** and we recommend reading it before applying. It will tell you quickly whether this work interests you.

We are engaging physicists to work on research-level physics problems in their own subfield. Depending on where your publication record fits, that can mean creating problems, solving them, reviewing completed work, or auditing it. We agree the specific assignment with you once you are matched to an area.

This is research-grade work rather than volume work. Whatever you produce has to be complete enough for another specialist in your subfield to follow and verify independently, so written reasoning is part of every assignment.

## Research areas in this panel

Seven areas. We match narrowly: you need to have published on one of these specific phenomena, not in high energy theory broadly. Each area lists the methods it requires.

**1. AdS/BCFT, end-of-the-world branes, black hole interiors:** AdS/BCFT correspondence, Karch-Randall end-of-the-world branes, geodesic approximation for heavy bulk fields, thermal one-point functions, BTZ black hole geometry, black hole interior behind the horizon.

**2. LaMET and quasi-PDFs: matching kernels and DGLAP evolution:** Large-momentum effective theory, quasi-parton distribution functions, pion parton distribution functions, perturbative matching kernels, modified minimal subtraction renormalization, plus distributions, DGLAP evolution, renormalization-group resummation of logarithms, one-loop running coupling.

**3. LaMET and quasi-PDFs: one-loop perturbative QCD and renormalization:** Perturbative quantum chromodynamics, large-momentum effective theory, quasi-parton distribution functions, one-loop Feynman diagram computation, dimensional regularization, MS-bar renormalization scheme, superficial degree of divergence analysis, soft and collinear infrared divergences.

**4. LaMET and quasi-PDFs: Coulomb gauge matching and lattice QCD parton structure:** Large-momentum effective theory, Coulomb gauge fixing, quasi-parton distribution functions, perturbative matching kernels, one-loop QCD corrections, dimensional regularization, MS-bar renormalization scheme, lattice QCD calculations of parton structure.

**5. Ultralight vector (dark photon) dark matter in laser interferometers:** Ultralight vector dark matter, dark photon coupled to the baryon-minus-lepton current, laser interferometric gravitational-wave detection, differential test-mass forces from composition-dependent charge-to-mass ratios, strain noise spectral density and detector sensitivity curves, coherent signal integration and signal-to-noise scaling, local dark matter density normalization, coherence time of a stochastic ultralight field.

**6. Matrix models and invariant theory: Grassmann-valued invariants, Molien-Weyl and Hilbert series:** Adjoint matrix models, Grassmann-valued matrix invariants, gauge-invariant operator counting, trace relations, Cayley-Hamilton identities, Molien-Weyl formula, plethystic Hilbert series, primary and secondary invariants.

**7. Superconformal and Witten index, finite-N trace relations, plethystic counting:** Witten index, superconformal index, adjoint fermion matrix models, trace relations at finite N, Grassmann-valued matrix invariants, Molien-Weyl formula, plethystic counting of gauge invariants, finite-rank corrections to large-N counting.

## Methods we expect to find in your own publications

You should be able to point to your own papers demonstrating **at least one** of the following families:

- Holographic: AdS/CFT and AdS/BCFT correspondence, geodesic approximation, thermal one-point functions
- Perturbative QCD: one-loop Feynman diagram computation, dimensional regularization, MS-bar renormalization, soft and collinear divergences
- Effective field theory: large-momentum effective theory, perturbative matching kernels, renormalization-group resummation of logarithms
- Lattice: lattice QCD calculations of parton structure, gauge fixing
- Group-theoretic: gauge-invariant operator counting, trace relations, Cayley-Hamilton identities, Molien-Weyl formula, plethystic counting
- Astroparticle and detector: interferometric gravitational-wave detection, strain noise spectral density, signal-to-noise scaling

## Who we are looking for

**A PhD in high energy theory, nuclear theory or a closely related field.** This is a hard requirement. Postdoctoral researchers, research scientists and junior faculty are the strongest fit. Senior PhD students with a strong first-author record are welcome to apply.

**Published work on the specific phenomenon above, not the adjacent one.** This is the single most common reason we decline otherwise excellent physicists. Command of the methods is not enough if you have not published on the phenomenon itself.

For the quasi-PDF areas specifically, we need **loop-calculation experience, not lattice experience alone.**

**A verifiable publication record.** Three to five representative papers with arXiv IDs or DOIs, ideally from the last five years. First author strongly preferred. Every paper you list will be checked against the public record.

**Working proficiency with LaTeX, Python, SymPy and Jupyter.** Some familiarity with an agentic coding extension in VS Code is useful. Gaps here are acceptable if you declare them honestly.

**English at B2 or above**, including written reasoning. A large part of the value you add is how clearly you set out your argument.

## Application steps

1. **Apply and complete the attached form.** Basic information, education, research experience, your method self-attestation, and up to five of the areas above that you are the best fit for. For each area you select, give an arXiv ID or DOI of **your own paper** as proof, with your author position and the methods it demonstrates. A selection without proof is not scored.
2. **We verify your papers and authorship** against the public record.
3. **Then one of two things happens.** Either we onboard you directly, or we invite you to a short live alignment call to agree the area and the assignment with you.
4. A brief **30 to 45 minute assessment** may be added, but only where we need it. Most applicants will not see one.

## Commitment and rate

**10 hours per week**, sustained across an **8 to 10 week** window, starting immediately. Remote and asynchronous with no fixed hours.

**$80 to $110 per hour**, set by depth of subdomain match.

High Energy Physicist (PhD)

## About the work

**CritPt** is a public benchmark of research-level physics challenges, built to test whether frontier AI models can carry out genuine physics research reasoning rather than textbook problem solving. The benchmark paper is **[arXiv:2509.26574](https://arxiv.org/abs/2509.26574)** and we recommend reading it before applying. It will tell you quickly whether this work interests you.

We are engaging physicists to work on research-level physics problems in their own subfield. Depending on where your publication record fits, that can mean creating problems, solving them, reviewing completed work, or auditing it. We agree the specific assignment with you once you are matched to an area.

This is research-grade work rather than volume work. Whatever you produce has to be complete enough for another specialist in your subfield to follow and verify independently, so written reasoning is part of every assignment.

## Research areas in this panel

Twelve areas. We match narrowly: you need to have published on one of these specific phenomena, not in quantum information broadly. Each area lists the methods it requires.

**1. Distributed quantum sensing and metrology with GHZ probes under noise:** Distributed quantum sensing, quantum Fisher information, Greenberger-Horne-Zeilinger entangled probe states, GHZ-diagonal noisy states, Lindblad master equations, multiparameter phase estimation, entanglement distribution in quantum networks.

**2. Quantum optics: two-mode squeezing, SU(1,1) interferometry, homodyne detection with loss:** Optical parametric amplification, two-mode squeezed vacuum, two-photon formalism for sideband modes, Bogoliubov transformation of field operators, SU(1,1) nonlinear interferometry, cascaded phase-sensitive amplification with intermediate loss, balanced homodyne detection of quadrature sidebands, beam-splitter model of optical loss and detection inefficiency.

**3. Quantum Shannon theory: Holevo and accessible information, classical capacity:** Holevo information, accessible information of quantum ensembles, classical capacity of quantum channels, classical-quantum states, von Neumann entropy, optimality conditions for information-maximizing ensembles, convex optimization over probability distributions.

**4. Holographic codes and random tensor networks, Haar averaging over O(N):** Non-isometric holographic codes, Haar averaging over the orthogonal group, Weingarten calculus, random tensor network models of holography, moments of random linear maps, post-selection onto fiducial states, black hole interior reconstruction.

**5. Matrix product states, sequential generation, holographic quantum simulation:** Matrix product states, finitely correlated states, sequential generation of matrix product states, isometric tensor network circuits, transfer-matrix formalism for correlation functions, exponential decay of correlations, holographic quantum simulation, bond dimension and entanglement structure.

**6. Eigenstate-to-Hamiltonian construction, commutant algebras, integrals of motion:** Eigenstate-to-Hamiltonian construction, symmetric Hamiltonian construction from integrals of motion, quantum covariance matrix method, commutant algebras of local Hamiltonian families, Pauli string operator bases, Jordan-Wigner transformation, exact diagonalization of spin chains.

**7. Quantum-limited imaging and superresolution: QFI, spatial-mode demultiplexing:** Quantum Fisher information, quantum Cramer-Rao bound, Sudarshan-Glauber P representation, mutual coherence matrix of thermal sources, point spread function and diffraction-limited imaging, weak-source single-photon regime, spatial-mode demultiplexing, direct imaging and the Rayleigh resolution limit.

**8. Quantum walks and spatial search on graphs, degenerate perturbation theory:** Continuous-time quantum walk, spatial search on graphs, graph adjacency matrix spectra, simplex of complete graphs and truncated simplex lattices, degenerate perturbation theory, symmetry-reduced invariant subspaces, critical jumping rate and avoided level crossing.

**9. Quantum foundations: higher-order interference, Sorkin parameter, GPTs as information resource:** Higher-order interference, Sorkin parameter, multipath single-particle interferometry, spatial superposition as an information resource, generalized probabilistic theories, classical correlation polytopes, fingerprinting inequality, single-particle multiple access channels.

**10. Quantum foundations: generalized noncontextuality, KCBS, ontological models:** Generalized noncontextuality, Kochen-Specker contextuality, KCBS inequality, noncontextual ontological models, generalized probabilistic theories, simplex embedding, linear programming for nonclassicality, unsharp measurements and POVMs, white-noise robustness.

**11. Quantum Shannon theory: private states, quantum and private capacity, superactivation:** Private states and private bits, Choi-Jamiolkowski channel-state duality, quantum capacity and coherent information, private capacity of quantum channels, Werner states and symmetric subspace projections, degradable and antidegradable channels, superadditivity and superactivation of quantum capacity, spin alignment conjecture.

**12. Quantum relative entropy, strong data processing, contraction coefficients of qubit channels:** Quantum relative entropy, data processing inequality for relative entropy, contraction coefficients of quantum channels, strong data processing inequalities, qubit amplitude damping channel, monotone Riemannian metrics and the Bogoliubov-Kubo-Mori metric, Bloch-sphere representation of qubit channels, trace-distance contraction bounds.

## Methods we expect to find in your own publications

You should be able to point to your own papers demonstrating **at least one** of the following families:

- Entropic and information-theoretic: von Neumann entropy, Holevo bound, coherent information, contraction coefficients, Choi-Jamiolkowski duality
- Open quantum systems: Lindblad master equations, quantum channels, degradable and antidegradable channels, amplitude damping and dephasing models
- Random matrix and Haar integration: Weingarten calculus, moments of random linear maps, Haar averaging over the orthogonal group
- Tensor network: matrix product states, transfer-matrix formalism, isometric tensor network circuits, exact diagonalization of spin chains
- Convex and geometric: convex optimization over probability distributions, simplex embedding, correlation polytopes, linear programming for nonclassicality
- Estimation theory: quantum Fisher information, quantum Cramer-Rao bounds, multiparameter phase estimation

## Who we are looking for

**A PhD in quantum information, quantum optics or a closely related field.** This is a hard requirement. Postdoctoral researchers, research scientists and junior faculty are the strongest fit. Senior PhD students with a strong first-author record are welcome to apply.

**Published work on the specific phenomenon above, not the adjacent one.** This is the single most common reason we decline otherwise excellent physicists. Having applied the information-theoretic toolkit elsewhere is not enough if your papers are not on the phenomenon in question.

**A verifiable publication record.** Three to five representative papers with arXiv IDs or DOIs, ideally from the last five years. First author strongly preferred. Every paper you list will be checked against the public record.

**Working proficiency with LaTeX, Python, SymPy and Jupyter.** Some familiarity with an agentic coding extension in VS Code is useful. Gaps here are acceptable if you declare them honestly.

**English at B2 or above**, including written reasoning. A large part of the value you add is how clearly you set out your argument.

## Application steps

1. **Apply and complete the attached form.** Basic information, education, research experience, your method self-attestation, and up to five of the areas above that you are the best fit for. For each area you select, give an arXiv ID or DOI of **your own paper** as proof, with your author position and the methods it demonstrates. A selection without proof is not scored.
2. **We verify your papers and authorship** against the public record.
3. **Then one of two things happens.** Either we onboard you directly, or we invite you to a short live alignment call to agree the area and the assignment with you.
4. A brief **30 to 45 minute assessment** may be added, but only where we need it. Most applicants will not see one.

## Commitment and rate

**10 hours per week**, sustained across an **8 to 10 week** window, starting immediately. Remote and asynchronous with no fixed hours.

**$80 to $110 per hour**, set by depth of subdomain match.

Quantum Information Science Physicist (PhD)

## About the work

**CritPt** is a public benchmark of research-level physics challenges, built to test whether frontier AI models can carry out genuine physics research reasoning rather than textbook problem solving. The benchmark paper is **[arXiv:2509.26574](https://arxiv.org/abs/2509.26574)** and we recommend reading it before applying. It will tell you quickly whether this work interests you.

We are engaging physicists to work on research-level physics problems in their own subfield. Depending on where your publication record fits, that can mean creating problems, solving them, reviewing completed work, or auditing it. We agree the specific assignment with you once you are matched to an area.

This is research-grade work rather than volume work. Whatever you produce has to be complete enough for another specialist in your subfield to follow and verify independently, so written reasoning is part of every assignment.

## Research areas in this panel

Nineteen areas. We match narrowly: you need to have published on one of these specific phenomena, not in condensed matter broadly. Each area lists the methods it requires.

**1. One-dimensional field theory: bosonization, Majorana and Ising-Luttinger sectors, RG:** Abelian bosonization, Majorana fermion field theory in one dimension, coupled Ising and Luttinger-liquid sectors, scaling dimensions of vertex operators, one-loop renormalization-group flow equations, operator product expansion, commensurate-incommensurate transition.

**2. Parafermion zero modes, non-Abelian braiding, FQH-superconductor heterostructures:** Parafermion zero modes, ZN parafermion algebra, non-Abelian braiding statistics, fusion channels of topological defects, fractional Josephson effect, adiabatic evolution and Berry phases, fractional quantum Hall-superconductor heterostructures.

**3. Rational CFT, Verlinde lines, non-Abelian quantum Hall edge theories:** Rational conformal field theory on the torus, Verlinde lines as topological defect lines, modular S matrix and the Verlinde formula, Moore-Read non-Abelian quantum Hall state, chiral edge theories of quantum Hall states, fusion categories and non-invertible symmetries, modular invariance of the primary field spectrum.

**4. SPT and Haldane phase, matrix product states, decoherence of topological order:** AKLT valence-bond-solid model, Haldane phase and symmetry-protected topological order, string order parameters, matrix product states, transfer-matrix evaluation of correlation functions, Kraus-operator quantum channels, decoherence-induced mixed-state topological order.

**5. Flat-band Hubbard models, flat-band ferromagnetism, Hartree-Fock and DQMC:** Checkerboard-lattice Hubbard model, topological nearly flat bands, flat-band ferromagnetism, Hartree-Fock mean-field theory, Stoner ferromagnetic instability, interaction-driven quantum phase transitions, determinant quantum Monte Carlo.

**6. SYK model, large-N disordered fermions, Schwinger-Dyson thermodynamics:** Sachdev-Ye-Kitaev model, low-rank random couplings, Majorana fermions, quenched disorder averaging, large-N melonic diagrammatics, Schwinger-Dyson equations, thermodynamic free energy, residual zero-temperature entropy.

**7. Correlated electron transport: Hubbard model, Kubo formalism, diagrammatic perturbation theory:** Hubbard model, Fermi-liquid theory, Kubo linear-response formalism, perturbative diagrammatic expansion in the interaction, Matsubara Green's function technique, vertex corrections and momentum relaxation, electron-electron scattering on a lattice, low-density expansion near a band edge.

**8. Wigner crystallization, two-component Coulomb systems, quantum melting:** Wigner crystallization, two-component Coulomb crystallization, critical mass ratio for quantum crystals, electron-hole bilayers, Wigner-Seitz coupling parameter, Lindemann criterion for quantum melting, dimensional analysis of competing energy scales, quantum Monte Carlo phase diagrams.

**9. Quantum geometry, Wannier obstruction, Z2 topological invariants (Kane-Mele and Wilson loop):** Quantum geometric tensor, quantum metric, gauge-invariant Wannier spread, maximally localized Wannier functions, Kane-Mele time-reversal Z2 invariant, Wilson loop and Wannier charge center winding, continuum models with plane-wave expansion, Wannier obstruction and exponential localization, Rashba spin-orbit coupling.

**10. Hatsugai-Kohmoto model, Mott physics, quasiparticle scattering and RG stability:** Hatsugai-Kohmoto model, Mott insulators and Hubbard bands, Fermi liquid theory, quasiparticle scattering rates and lifetimes, phase-space constraints on fermion-fermion scattering, renormalization-group stability of interacting fixed points.

**11. Schwinger-Keldysh EFT, dissipative hydrodynamics, fracton and multipole symmetries:** Schwinger-Keldysh effective field theory, dissipative hydrodynamics, multipole and dipole symmetries, fracton phases of matter, spontaneous symmetry breaking and Goldstone stiffness, superfluid stiffness and susceptibilities, derivative expansion power counting, hydrodynamic mode dispersion relations.

**12. Quantum Monte Carlo for the 2D electron gas: DMC, finite-size corrections:** Diffusion Monte Carlo, Slater-Jastrow trial wave functions, two-dimensional finite-size corrections and thermodynamic-limit extrapolation, periodic boundary conditions with long-range Coulomb interactions, static structure factor and pair correlation functions, random phase approximation for long-wavelength fluctuations, Fermi liquid theory.

**13. Dirac materials transport: Coulomb disorder, nonlinear screening, Boltzmann transport:** Dirac fermions in graphene, charged impurity Coulomb disorder, nonlinear screening and Thomas-Fermi theory, electron-hole puddles, self-consistent transport theory, topological insulator surface states, long-range versus short-range scattering, Boltzmann transport and mean free path.

**14. Topological crystalline insulators: disclinations, fractional charge, filling anomaly:** Topological crystalline insulators, band representations and Wyckoff positions, Wannier orbital angular momentum under rotation symmetry, disclination defects and Frank angle, translation-equivalence classes of lattice defects, fractional charge quantization, filling anomaly, discrete shift invariant.

**15. Kitaev honeycomb model, Z2 gauge theory, exactly solvable spin liquids:** Kitaev honeycomb model, Majorana fermionization of spins, Z2 lattice gauge theory and flux sectors, exactly solvable quantum spin liquids, fermion parity projection onto physical states, topological ground-state degeneracy on a torus, Bogoliubov-de Gennes diagonalization, exact diagonalization of finite lattices.

**16. Semiclassical thermoelectric transport: anisotropic Seebeck, bipolar conduction:** Semiclassical Boltzmann transport theory, relaxation-time approximation, anisotropic effective-mass tensors, Seebeck coefficient and thermopower, bipolar transport in intrinsic semiconductors, axis-dependent conduction polarity, two-band electronic structure models.

**17. Inelastic neutron scattering theory: multiphonon expansion, dynamic structure factor:** Inelastic neutron scattering, multiphonon expansion of the scattering law, quantum harmonic oscillator response, incoherent dynamic structure factor, Debye-Waller factor, bound-atom scattering cross section, low-temperature limit of the scattering law, neutron total cross section.

**18. Magnetic space groups, neutron-diffraction magnetic structure, magneto-optics:** Magnetic space groups, Belov-Neronova-Smirnova notation, Wyckoff position analysis, magnetic propagation vectors from neutron diffraction, multi-k magnetic structures, irreducible representation analysis of magnetic order, time-reversal symmetry breaking, symmetry-adapted tensor properties, magneto-optic Kerr effect.

**19. Crystallography of modulated structures: kinematic diffraction, superspace symmetry:** Kinematic diffraction theory, crystal structure factor, displacively modulated crystal structures, satellite and superlattice reflections, reciprocal lattice and Brillouin zone indexing, diffraction selection rules and extinction conditions, superspace symmetry of modulated crystals, longitudinal static strain waves in crystals.

## Methods we expect to find in your own publications

You should be able to point to your own papers demonstrating **at least one** of the following families:

- Field-theoretic: bosonization, conformal field theory on the torus, Schwinger-Keldysh effective field theory, renormalization-group flow
- Many-body numerics: quantum Monte Carlo (diffusion and determinant), exact diagonalization, matrix product states, transfer-matrix methods
- Band theory and topology: Wannier functions, Berry phases, Wilson loops, Wyckoff positions, band representations
- Response and transport: Kubo linear response, Matsubara Green's functions, diagrammatic perturbation theory, Boltzmann transport
- Scattering and symmetry analysis: neutron scattering theory, magnetic space groups, irreducible representation analysis, kinematic diffraction

## Who we are looking for

**A PhD in condensed matter physics or a closely related field.** This is a hard requirement. Postdoctoral researchers, research scientists and junior faculty are the strongest fit. Senior PhD students with a strong first-author record are welcome to apply.

**Published work on the specific phenomenon above, not the adjacent one.** This is the single most common reason we decline otherwise excellent physicists. Publishing on topological order generally is not enough if your papers are not on the particular model in question. Command of the methods is not enough if you have not published on the phenomenon itself.

**A verifiable publication record.** Three to five representative papers with arXiv IDs or DOIs, ideally from the last five years. First author strongly preferred. Every paper you list will be checked against the public record.

**Working proficiency with LaTeX, Python, SymPy and Jupyter.** Some familiarity with an agentic coding extension in VS Code is useful. Gaps here are acceptable if you declare them honestly.

**English at B2 or above**, including written reasoning. A large part of the value you add is how clearly you set out your argument.

## Application steps

1. **Apply and complete the attached form.** Basic information, education, research experience, your method self-attestation, and up to five of the areas above that you are the best fit for. For each area you select, give an arXiv ID or DOI of **your own paper** as proof, with your author position and the methods it demonstrates. A selection without proof is not scored.
2. **We verify your papers and authorship** against the public record.
3. **Then one of two things happens.** Either we onboard you directly, or we invite you to a short live alignment call to agree the area and the assignment with you.
4. A brief **30 to 45 minute assessment** may be added, but only where we need it. Most applicants will not see one.

## Commitment and rate

**10 hours per week**, sustained across an **8 to 10 week** window, starting immediately. Remote and asynchronous with no fixed hours.

**$80 to $110 per hour**, set by depth of subdomain match.

Condensed Matter Physicist (PhD)

Mercor is partnering with a leading AI research lab to build next-generation text-to-speech (TTS) systems capable of producing natural, expressive, and human-like voices. We are seeking **South Africa-based native English speakers with authentic South African accents** to contribute high-quality voice recordings for training and evaluating cutting-edge speech models. This role is open to both experienced voice professionals and newcomers with a naturally clear, expressive voice and a good recording setup — prior voice-acting experience is a plus but not required.

**This is an ongoing opening — we onboard new voices on a rolling basis, so applications are always welcome. The engagement itself is a single recording session of approximately 4 hours. A second session of similar length may follow depending on the client's needs, though this isn't guaranteed.**

**Your voice recordings will be used exclusively for the client's internal CX AI Agent. They will not be sold, licensed, or reused for any other product, dataset, or purpose.**

* * *

### **Key Responsibilities**

- Record high-quality voice samples across diverse scripts (conversational, narrative, instructional, etc.)
    
- Deliver **clear, natural, and expressive speech** with strong control over tone, pacing, and pronunciation
    
- Maintain **consistency in voice, accent, and delivery** across recording sessions
    
- Follow detailed recording guidelines (environment, microphone setup, file formatting)
    
- Perform multiple takes with variation in emotion, emphasis, and style when required
    

### **Requirements**

- **Native English speaker with an authentic South African accent, currently based in South Africa**
    
- A clear, natural, and expressive voice (prior voice acting, narration, or broadcast experience is a plus, not required)
    
- Access to a **professional or near-professional recording setup** (quality microphone, quiet environment, pop filter, etc.)
    
- Strong command of **intonation, diction, and emotional range**
    
- Ability to follow scripts precisely while maintaining natural delivery
    
- Availability for **a single ~4-hour recording session**, with the possibility of a further session of similar length later on
    
- **\[IMP\]: Your voice may be cloned for the client's CX AI Agent so please only apply if you are okay with voice cloning. Your recordings will not be sold or reused outside of this specific use case.**
    

### **Preferred Qualifications**

- Experience recording for **TTS, audiobooks, IVR systems, or AI voice datasets**
    
- Familiarity with audio editing tools (e.g., Audacity, Adobe Audition, Reaper)
    
- Ability to deliver **multiple vocal styles** (e.g., conversational, corporate, energetic, calm)
    

* * *

Voice Actor: CX Agent Voice Cloning (South African English)

## About the work

Mercor is designing a benchmark that measures how well AI agents do real enterprise cyber defense work. Before we build it, we want to hear from people who do that work every day.

This is a short paid research engagement, not a task or data project. You will join two to three one-hour video conversations with the Mercor team over roughly four weeks, scheduled around your availability.

## What we will talk about

- How your work actually happens: the tools you live in, what lands in your queue, where judgment matters and where it does not.
- How your team measures whether a triage, investigation, containment, hunt, detection rule or patch was done well.
- Where AI assistance helps you today, where it gets in the way, and what would make a benchmark of AI on this work credible to you.
- Our draft task taxonomy: what is missing, what is mislabeled, what you would weight differently.

## Who we are looking for

Current or recent hands-on practitioners in enterprise security, for example:

- SOC analysts and SOC leads (tier 2 and above)
- Incident responders and digital forensics investigators
- Detection engineers and threat hunters
- Application security or product security engineers who find and fix vulnerabilities in production code
- Security engineers who have run or been on the receiving end of a red team engagement

Experience with an EDR and SIEM stack (CrowdStrike Falcon, Microsoft Defender and Sentinel, Splunk, SentinelOne, Elastic or similar) is a plus. Security leaders are welcome if you are still close to the work.

## Pay and time

- Paid hourly for interview time as a spot bonus. Sessions are one hour; no prep is expected.
- Two to three sessions in total. Nothing to build, label or submit.
- Remote, US-based.

## Confidentiality

We will ask you to describe patterns, not specific incidents, and to leave out anything confidential about current or former employers or customers. Your input is used only to design the benchmark and is never attributed to you or your employer.

Cybersecurity Practitioner: Paid Expert Interviews (SOC, Incident Response, Detection, AppSec)

**Join a product engineering team building native mobile and full-stack experiences used by customers every day.**

# 1\. Overview

We're partnering with a large enterprise technology organization to place a Staff Software Engineer on the team behind its Android and iOS apps. It's a mobile-first role, but engineers here also pitch in on backend services, APIs, and web, and use AI coding tools as a normal part of their day. If you enjoy owning features end to end and learning new parts of the stack, this is a great fit. This is a full-time commitment of 40 hours per week.

This is a W-2 employment position with Cincinnatus LLC, with the opportunity to be placed within a leading enterprise client's engineering organization as part of their extended workforce.

# 2\. Key Responsibilities

- Build and ship features for the team's native mobile apps, mainly Android in Kotlin.
    
- Contribute beyond mobile when needed, across backend services, APIs, and web applications.
    
- Use AI-powered developer tools to get up to speed on unfamiliar code and ship with quality.
    
- Take part in technical design, code reviews, and production support.
    
- Work closely with Product, UX, and partner engineering teams, and help other engineers grow.
    

# 3\. Core Qualifications

- 4+ years of professional software engineering experience.
    
- Hands-on native mobile experience, ideally Android with Kotlin and Jetpack. iOS/Swift is welcome too, and Flutter or React Native experience counts if you've also done some native work.
    
- Some production experience outside mobile, such as backend services, APIs, or web apps.
    
- Solid engineering fundamentals: design, testing, and debugging.
    
- Comfortable using AI coding tools such as Copilot, Cursor, or Claude.
    
- Ability to engage reliably for at least 40 hours/week during weekdays.
    
- Clear communication and a collaborative, curious mindset.
    

**Nice to have:** exposure to LLM-powered features (RAG, embeddings, agents), experience with accessibility standards, or work on apps with accounts, payments, or orders. You don't need all of these to apply.

**About Cincinnatus LLC:**
Cincinnatus LLC is an enterprise staffing company that partners with leading technology companies to source and employ highly skilled professionals for contingent and contract-based opportunities. Cincinnatus serves as the employer of record for these engagements, providing W-2 employment, payroll, benefits, and compliance, while placing employees directly within client teams to work on high-impact initiatives.

**Equal Employment Opportunity:**
Cincinnatus is proud to be an Equal Employment Opportunity employer. We do not discriminate based upon race, religion, color, national origin, sex (including pregnancy, childbirth, reproductive health decisions, or related medical conditions), sexual orientation, gender identity, gender expression, age, status as a protected veteran, status as an individual with a disability, genetic information, political views or activity, or any other legally protected characteristic.

Staff Software Engineer, Mobile (Android, Kotlin, Full Stack)

**Join a high-scale consumer platform team as a Senior Backend Engineer, building the services behind digital experiences used by millions of people every day.**

# 1\. Overview

We're partnering with a large enterprise technology organization to place a Senior Software Engineer on a platform team that powers its core customer-facing experiences across web and mobile. The team works across Engineering, UX, Product, Business, and Analytics to ship fast, iterate through continuous experimentation and customer feedback, and integrate advanced capabilities such as Generative AI into production services. We're seeking senior backend engineers with hands-on Node.js, TypeScript, and Rust experience and a deep understanding of how the web works at scale. This is a full-time commitment of 40 hours per week.

This is a W-2 employment position with Cincinnatus LLC, with the opportunity to be placed within a leading enterprise client's engineering organization as part of their extended workforce.

# 2\. Key Responsibilities

- Lead delivery of scoped features through collaborative engineering teams, keeping work aligned with platform and system objectives.
    
- Design, develop, test, and maintain high-quality, efficient services in Node.js, TypeScript, and Rust, across the full software lifecycle and CI/CD.
    
- Build and optimize server-to-server integrations and API layers (REST and GraphQL) that serve high-traffic, customer-facing applications.
    
- Integrate AI/ML capabilities into product services and use generative AI tools to improve engineering productivity.
    
- Contribute to solution design, conduct code reviews, and apply best practices for scalable, secure, and maintainable systems.
    
- Monitor application performance, support defect resolution, and drive continuous improvement.
    
- Mentor peers and contribute to knowledge sharing and engineering excellence across the team.
    

# 3\. Core Qualifications

- 4+ years of professional software engineering experience (or a Bachelor's degree in Computer Science or a related field and 3+ years), building production systems at a recognized, high-scale organization.
    
- Hands-on production coding experience in Node.js and TypeScript on the backend.
    
- Hands-on coding experience in Rust, in production services, performance-critical modules, or substantial systems work.
    
- Strong web fundamentals: HTTP semantics, cookies, headers, caching, sessions, and server-to-server communication patterns (auth between services, timeouts, retries).
    
- Experience building or operating high-traffic, consumer-facing web platforms.
    
- Experience with GraphQL on Node.js (schema design, resolvers, federation) is a plus.
    
- Familiarity with web accessibility standards is a plus.
    
- Demonstrable career progression to senior-level ownership.
    
- Ability to engage reliably for at least 40 hours/week during weekdays.
    
- Excellent written and verbal communication skills, and the ability to explain complex technical concepts clearly to broad audiences.
    

**About Cincinnatus LLC:**
Cincinnatus LLC is an enterprise staffing company that partners with leading technology companies to source and employ highly skilled professionals for contingent and contract-based opportunities. Cincinnatus serves as the employer of record for these engagements, providing W-2 employment, payroll, benefits, and compliance, while placing employees directly within client teams to work on high-impact initiatives.

**Equal Employment Opportunity:**
Cincinnatus is proud to be an Equal Employment Opportunity employer. We do not discriminate based upon race, religion, color, national origin, sex (including pregnancy, childbirth, reproductive health decisions, or related medical conditions), sexual orientation, gender identity, gender expression, age, status as a protected veteran, status as an individual with a disability, genetic information, political views or activity, or any other legally protected characteristic.

Senior Software Engineer, Backend (Node.js, TypeScript, Rust)

## About the work

We are building expert-authored forecasting and research data used to evaluate and improve AI models on real political and financial analysis. Each task asks a practitioner to reason from a fixed evidence cutoff — no hindsight — and document the judgment a strong analyst would have made at that moment.

This role covers political report authorship: elections, polling, and political-risk events.

## What you will do

- Author point-in-time political reports and forecasts from a stated evidence cutoff
    
- Make and document directional judgments about polling, win probabilities, and market-implied odds
    
- Explain the causal mechanism behind an event's expected political and market impact
    
- Review model-generated analyses and grade them against your own standard
    

## Who we are looking for

- Local political analysts with hands-on experience on specific races in swing states: current or upcoming Senate, governor, or statewide contests, or recent cycles in the same state
    
- On-the-ground knowledge of the electorate, the candidates, and the local political environment, from roles such as university political scientist, state or regional pollster, state-level campaign or party analytics staff, or state political-risk analyst
    
- A record of making live judgments about polling, win probabilities, or odds in those races — not retrospective commentary
    
- Strong polling literacy and causal event analysis
    
- Typically 8+ years of direct experience, or unusually strong evidence of equivalent performance
    

Backgrounds we look at first: political science departments and survey research centers at universities in the target states, state and regional pollsters, state-level campaign and party analytics teams, and decision desks or race-ratings outlets with state-specific coverage. National forecasting or political-risk experience is welcome when paired with deep knowledge of a specific state. Brand-name experience is a strong first-pass signal but is not a hard requirement — direct experience and performance on the work sample determine who qualifies.

## Not a fit

- Profiles whose experience is mainly international politics, geopolitics, or country risk, with little or no work on US races
    
- Analysts who have not worked on a current or recent US race in the state they would cover
    
- Pure communications or journalism profiles without demonstrated forecasting, polling, or probabilistic judgment
    
- Academic profiles whose work is purely theoretical, with no analysis of live races
    

## Point-in-time discipline

Every task is graded on reasoning from the stated cutoff without hindsight, data leakage, confidential information, or material non-public information.

Local Political Expert — Elections Forecasting & Political Risk

**About the Role**

Mercor is partnering with a leading AI lab to train frontier models on high-quality insurance reasoning data. We're hiring **Insurance** **Actuarial Specialists** to design realistic pricing, reserving, and risk-modeling scenarios, evaluate model outputs against established actuarial standards, and help shape how the next generation of AI reasons about probability, loss estimation, and financial risk.

We welcome actuaries across all practice areas, including life, health, property & casualty, pensions/retirement, and enterprise risk — from insurers, reinsurers, consultancies, and corporate actuarial teams.

**What You'll Do**

- Design realistic scenarios involving ratemaking, loss reserving, valuation, experience studies, assumption-setting, and profitability analysis
    
- Create work products such as pricing models, reserve reviews, actuarial memoranda, technical assumptions, and results summaries
    
- Write "golden" reference responses at experienced actuarial-professional quality
    
- Grade AI-generated responses against structured rubrics for methodological soundness, calculation accuracy, assumption justification, and clear communication
    
- Identify flawed methods, calculation errors, unsupported assumptions, and gaps in documented rationale
    
- Provide written feedback the research team uses to improve model behavior
    
- Participate in onboarding office hours and calibration sessions
    

**You're a Good Fit If You**

- Have 2+ years of professional actuarial experience (any practice area)
    
- Understand how methods, assumptions, and data drive pricing, reserves, and valuation
    
- Can select an appropriate method, run the analysis, and articulate a defensible conclusion with a documented rationale
    
- Demonstrate excellent quantitative analysis, balanced judgment, and high attention to detail
    
- Can distinguish actuarial analysis from underwriting decisions and from formal financial or legal advice
    

**Bonus Qualifications**

- ASA, FSA, ACAS, FCAS, CERA, MAAA, or comparable credential (or actively passing exams)
    
- Experience across multiple practice areas or specialty lines
    
- Reinsurance, capital modeling, or enterprise risk experience
    
- Proficiency with R, Python, SAS, SQL, or actuarial software (Prophet, Emblem, Radar, ResQ)
    
- Experience with regulatory filings, rate filings, or statements of actuarial opinion
    

**Role Highlights**

- Minimum 20 hours per week (ideally 40+)
    
- Role starts immediately, applications reviewed on a rolling basis

Insurance Actuarial Specialists

## About the role

We are hiring expert Evaluators in **Software Engineering** to review and assess AI-generated work products (slides, spreadsheets, and documents) for real-world quality. You will apply deep subject-matter expertise to grade outputs.

This is a remote, hourly engagement.

## Requirements (must have)

1. **5+ years of relevant professional experience in Software Engineering at a top firm in the United States, United Kingdom, Canada, Australia, or New Zealand.**
    
2. **Native or professional fluency in English.**
    
3. **Highly proficient in Microsoft Office and Google Workspace** (PowerPoint, Excel, Word).
    

## Preferred (nice to have)

- Advanced degree (Master's or higher) from a reputable institution.
    

## What you'll do

- Evaluate AI-generated artifacts for usability in software engineering context.
    
- Identify factual, aesthetic, and presentation errors.
    
- Provide clear, structured written feedback.

Software Engineer Expert

## About the role

We are hiring expert Evaluators in **Data Science** to review and assess AI-generated work products (slides, spreadsheets, and documents) for real-world quality. You will apply deep subject-matter expertise to grade outputs.

This is a remote, hourly engagement.

## Requirements (must have)

1. **5+ years of relevant professional experience in Data Science at a top firm in the United States, United Kingdom, Canada, Australia, or New Zealand.**
    
2. **Native or professional fluency in English.**
    
3. **Highly proficient in Microsoft Office and Google Workspace** (PowerPoint, Excel, Word).
    

## Preferred (nice to have)

- Advanced degree (Master's or higher) from a reputable institution.
    

## What you'll do

- Evaluate AI-generated artifacts for usability in data science context.
    
- Identify factual, aesthetic, and presentation errors.
    
- Provide clear, structured written feedback.

Data Science Expert

## About the role

We are hiring expert Evaluators in **Finance** to review and assess AI-generated work products (slides, spreadsheets, and documents) for real-world quality. You will apply deep subject-matter expertise to grade outputs.

This is a remote, hourly engagement.

## Requirements (must have)

1. **5+ years of relevant professional experience in Finance at a top firm in the United States, United Kingdom, Canada, Australia, or New Zealand.**
    
2. **Native or professional fluency in English.**
    
3. **Highly proficient in Microsoft Office and Google Workspace** (PowerPoint, Excel, Word).
    

## Preferred (nice to have)

- Advanced degree (Master's or higher) from a reputable institution.
    

## What you'll do

- Evaluate AI-generated artifacts for usability in finance context.
    
- Identify factual, aesthetic, and presentation errors.
    
- Provide clear, structured written feedback.

Finance Expert

**About the Role**

Mercor is partnering with a leading AI lab to train frontier models on high-quality insurance reasoning data. We're hiring **Insurance Underwriting Specialists** to design realistic underwriting and risk-selection scenarios, evaluate model outputs against established industry standards, and help shape how the next generation of AI reasons about risk assessment, policy provisions, pricing, and coverage decisions.

We welcome underwriters across all lines, including life, health, property, casualty, commercial, personal, specialty, marine, surety, and reinsurance — from carriers, MGAs, and reinsurers.

**What You'll Do**

- Design realistic scenarios involving risk selection, exposure analysis, policy terms and conditions, exclusions, endorsements, pricing, and coverage determinations
    
- Create work products such as underwriting decisions, risk assessments, rating and pricing rationales, coverage analyses, and referral write-ups
    
- Write "golden" reference responses at experienced underwriting-professional quality
    
- Grade AI-generated responses against structured rubrics for risk-selection accuracy, provision interpretation, pricing correctness, and sound judgment
    
- Identify misread policy language, flawed risk classifications, incorrect pricing math, and gaps in documented rationale
    
- Provide written feedback the research team uses to improve model behavior
    
- Participate in onboarding office hours and calibration sessions
    

**You're a Good Fit If You**

- Have 2+ years of professional experience in insurance underwriting (any line)
    
- Understand how policy provisions, exclusions, endorsements, and rating factors determine coverage and price
    
- Can weigh risk, exposure, and financial evidence and articulate a defensible decision with a documented rationale
    
- Demonstrate excellent written analysis, balanced judgment, and high attention to detail
    
- Can distinguish underwriting decisions from sales/agent recommendations and from formal legal advice
    

**Bonus Qualifications**

- CPCU, CLU, AINS, ARM, AU, FALU, or comparable industry credential
    
- Experience across multiple lines of business or specialty risks
    
- Reinsurance, treaty, or facultative underwriting experience
    
- Actuarial, pricing, or catastrophe-modeling exposure
    
- Experience with regulatory filings, compliance, or audit review
    

**Role Highlights**

- Minimum 20 hours per week (ideally 40+)
    
- Role starts immediately, applications reviewed on a rolling basis

Insurance Underwriting Specialists

## About the role

We are hiring expert Evaluators in **Management Consulting** to review and assess AI-generated work products (slides, spreadsheets, and documents) for real-world quality. You will apply deep subject-matter expertise to grade outputs.

This is a remote, hourly engagement.

## Requirements (must have)

1. **5+ years of relevant professional experience in Management Consulting at a top firm in the United States, United Kingdom, Canada, Australia, or New Zealand.**
    
2. **Native or professional fluency in English.**
    
3. **Highly proficient in Microsoft Office and Google Workspace** (PowerPoint, Excel, Word).
    

## Preferred (nice to have)

- Advanced degree (Master's or higher) from a reputable institution.
    

## What you'll do

- Evaluate AI-generated artifacts for usability in consulting context.
    
- Identify factual, aesthetic, and presentation errors.
    
- Provide clear, structured written feedback.

Management Consulting Expert

**How to Apply — Required First Step**

This application includes a short required form, the **ChatGPT Power User Interest Form**. It appears as a step inside the application itself — there is no external link to open.

**Applications submitted without this form completed will not be considered.** Please answer every question before you submit.

**About the Opportunity**

A leading AI research organization is launching a program with heavy ChatGPT users and is seeking experts based in Mexico who are willing to share their personal ChatGPT history to support model research and evaluation.

This opportunity has two phases. Phase one is a one-time, deliverable-based step: link/share your personal ChatGPT history and complete a short consent form. Experts who complete phase one may then be invited into phase two, which is ongoing project work on the same program. Phase two is not part of the data-sharing step.

**What's Involved**

- Complete a short consent form authorizing the research organization to access your personal ChatGPT history
- Link / share your ChatGPT history following the instructions provided
- Provide the email address associated with your personal ChatGPT account (this may differ from your Mercor email)
- Self-verify that your account meets the eligibility criteria below (step-by-step instructions will be provided)

**Who We're Looking For — Eligibility**

You must meet all of the following:

- Based in **Mexico**
- **Native Spanish speaker**
- Have **lived and studied in Mexico for at least 3–5 years**
- Your ChatGPT conversations **must be primarily in Spanish**
- A heavy ChatGPT user: **40+ conversations** over the past nine months
- At least **1 saved memory** on your account
- A **personal** ChatGPT account that is **more than 3 months old**
- Only **personal** ChatGPT accounts qualify. Work, employer, and enterprise accounts are out of scope for this program, as are any accounts holding proprietary or confidential material.

**Verification**

Eligibility is confirmed by the research organization after you complete the consent form. Verification may take up to two weeks.

Mexican ChatGPT User

**How to Apply — Required First Step**

This application includes a short required form, the **ChatGPT Power User Interest Form**. It appears as a step inside the application itself — there is no external link to open.

**Applications submitted without this form completed will not be considered.** Please answer every question before you submit.

**About the Opportunity**

A leading AI research organization is launching a program with heavy ChatGPT users and is seeking experts based in Indonesia who are willing to share their personal ChatGPT history to support model research and evaluation.

This opportunity has two phases. Phase one is a one-time, deliverable-based step: link/share your personal ChatGPT history and complete a short consent form. Experts who complete phase one may then be invited into phase two, which is ongoing project work on the same program. Phase two is not part of the data-sharing step.

**What's Involved**

- Complete a short consent form authorizing the research organization to access your personal ChatGPT history
- Link / share your ChatGPT history following the instructions provided
- Provide the email address associated with your personal ChatGPT account (this may differ from your Mercor email)
- Self-verify that your account meets the eligibility criteria below (step-by-step instructions will be provided)

**Who We're Looking For — Eligibility**

You must meet all of the following:

- Based in **Indonesia**
- **Native Bahasa Indonesia speaker**
- Have **lived and studied in Indonesia for at least 3–5 years**
- Your ChatGPT conversations **must be primarily in Bahasa Indonesia**
- A heavy ChatGPT user: **40+ conversations** over the past nine months
- At least **1 saved memory** on your account
- A **personal** ChatGPT account that is **more than 3 months old**
- Only **personal** ChatGPT accounts qualify. Work, employer, and enterprise accounts are out of scope for this program, as are any accounts holding proprietary or confidential material.

**Verification**

Eligibility is confirmed by the research organization after you complete the consent form. Verification may take up to two weeks.

Indonesian ChatGPT User

**How to Apply — Required First Step**

This application includes a short required form, the **ChatGPT Power User Interest Form**. It appears as a step inside the application itself — there is no external link to open.

**Applications submitted without this form completed will not be considered.** Please answer every question before you submit.

**About the Opportunity**

A leading AI research organization is launching a program with heavy ChatGPT users and is seeking experts based in South Korea who are willing to share their personal ChatGPT history to support model research and evaluation.

This opportunity has two phases. Phase one is a one-time, deliverable-based step: link/share your personal ChatGPT history and complete a short consent form. Experts who complete phase one may then be invited into phase two, which is ongoing project work on the same program. Phase two is not part of the data-sharing step.

**What's Involved**

- Complete a short consent form authorizing the research organization to access your personal ChatGPT history
- Link / share your ChatGPT history following the instructions provided
- Provide the email address associated with your personal ChatGPT account (this may differ from your Mercor email)
- Self-verify that your account meets the eligibility criteria below (step-by-step instructions will be provided)

**Who We're Looking For — Eligibility**

You must meet all of the following:

- Based in **South Korea**
- **Native Korean speaker**
- Have **lived and studied in South Korea for at least 3–5 years**
- Your ChatGPT conversations **must be primarily in Korean**
- A heavy ChatGPT user: **40+ conversations** over the past nine months
- At least **1 saved memory** on your account
- A **personal** ChatGPT account that is **more than 3 months old**
- Only **personal** ChatGPT accounts qualify. Work, employer, and enterprise accounts are out of scope for this program, as are any accounts holding proprietary or confidential material.

**Verification**

Eligibility is confirmed by the research organization after you complete the consent form. Verification may take up to two weeks.

Korean ChatGPT User

**How to Apply — Required First Step**

This application includes a short required form, the **ChatGPT Power User Interest Form**. It appears as a step inside the application itself — there is no external link to open.

**Applications submitted without this form completed will not be considered.** Please answer every question before you submit.

**About the Opportunity**

A leading AI research organization is launching a program with heavy ChatGPT users and is seeking experts based in **Japan** who are willing to share their personal ChatGPT history to support model research and evaluation.

This opportunity has **two phases**

- **Phase one** is a one-time, deliverable-based step: link/share your personal ChatGPT history and complete a short consent form.
    
- **Phase two:** Experts who complete phase one will then be invited into **phase two**, which is ongoing project work on the same program.
    

**What's Involved**

- Complete a short consent form authorizing the research organization to access your personal ChatGPT history
    
- Link / share your ChatGPT history following the instructions provided
    
- Provide the email address associated with your personal ChatGPT account (this may differ from your Mercor email)
    
- Self-verify that your account meets the eligibility criteria below (step-by-step instructions will be provided)
    

**Who We're Looking For — Eligibility**

You must meet all of the following:

- Based in **Japan**
    
- **Native Japan**
    
- Have **lived and studied in Japan for at least 3–5 years**
    
- Your ChatGPT conversations **must be primarily in Japan**
    
- A heavy ChatGPT user: **40+ conversations** over the past nine months
    
- At least **1 saved memory** on your account
    
- A **personal** ChatGPT account that is **more than 3 months old**
    
- Only **personal** ChatGPT accounts qualify. Work, employer, and enterprise accounts are out of scope for this program, as are any accounts holding proprietary or confidential material.
    

**Verification**

Eligibility is confirmed by the research organization after you complete the consent form. Verification may take up to two weeks.

Japanese ChatGPT User

We're hiring inpatient Registered Nurses (RNs) to support clinical AI product work — helping us develop and evaluate new products on an ongoing basis, which is why continuity and a reliable weekly commitment matter.

**What you'll do**

- Support annotation, clinical review, model evaluation, and user-feedback investigation
    
- Contribute to taxonomy/guideline development and other clinical product work as needs evolve
    
- Work directly with engineers, PMs, clinical scientists, and other clinicians
    

**What we're looking for**

- US based / able to practice in the US
    
- Recent inpatient bedside RN experience
    
- Broad clinical exposure: comfortable reviewing documentation across different inpatient populations, not overly specialized
    
- Self-starter / high ownership: can take an ambiguous clinical question, investigate it, and come back with a thoughtful recommendation
    
- Proactive and engaged: actively surfaces issues, patterns, and opportunities for improvement; comfortable speaking up when something doesn't seem right
    
- Rigorous and detail-oriented: able to consistently apply clinical logic and explain the rationale behind decisions
    
- Curious and adaptable: interested in learning how the product/model works and comfortable with processes that change quickly
    
- Strong communication and collaboration
    
- Able to reliably and consistently dedicate a minimum of 10 hours per week
    

**Preferred**

- Previous annotation experience with positive performance reviews
    
- Academic / large health system experience (e.g., a large teaching hospital, which exposes nurses to more complex patients, workflows, and documentation practices)

RN Annotators

## The role

We are hiring native and near-native Hindi speakers to **audit AI training data in their own language**, not to produce it.

You will work across two Amazon Sonic audit collections:

1. **Multilingual Transcription Audit.** You listen to a recording of human or agent speech, read the transcription an annotator produced, and judge whether it accurately captures what the speaker actually said. You apply a fixed set of error codes, assign a pass or fail verdict, and write a short rationale explaining the decision.
    
2. **Forced Alignment Audit.** You review audio that has already been segmented at the word level by an automated system. For each segment you verify that the start and end timestamps match where the word actually begins and ends, and that the reference transcription matches the spoken form. Where the machine got it wrong, you correct the boundary, split a segment, merge two, add a missing one, or remove one that contains no word.
    

Both collections run at 5.00 audit hours per task. Every judgment carries a written reason, so this is careful listening work rather than volume work.

## Who we are looking for

**Native or near-native command of Hindi as spoken in India.** You were either born and raised in a region where this language and variety is dominant, or you are fully fluent with five or more years of residence in such a region. This is a hard requirement.

**Strong English reading and writing.** All conventions, error codes, and audit rationale are written in English, so you need to read a detailed rulebook and write clear feedback in it.

**Disciplined rubric application.** The work rewards people who apply a written standard consistently across hundreds of judgments rather than relying on instinct. Prior transcription, subtitling, localization, linguistic annotation, or QA experience is a strong signal.

**Careful listening.** You can distinguish similar sounds, identify where one word ends and the next begins in continuous speech, and tell genuine acoustic ambiguity apart from a clear annotator error.

Helpful but not required: prior AI data annotation or model evaluation work, a phonetics or linguistics background, and experience with audio editing or waveform tools.

## What the work looks like

You listen, judge, and submit. Where you find an error you tag every applicable error code, not just the first one, and you explain in plain language why the task passed or failed. Critical errors are called out first so the reason for a failure is obvious to the reader.

You will complete a short calibration phase before production work begins.

Sonic Audit Specialist - Hindi

## The role

We are hiring native and near-native Mandarin Chinese speakers to **audit AI training data in their own language**, not to produce it.

You will work across two Amazon Sonic audit collections:

1. **Multilingual Transcription Audit.** You listen to a recording of human or agent speech, read the transcription an annotator produced, and judge whether it accurately captures what the speaker actually said. You apply a fixed set of error codes, assign a pass or fail verdict, and write a short rationale explaining the decision.

2. **Forced Alignment Audit.** You review audio that has already been segmented at the word level by an automated system. For each segment you verify that the start and end timestamps match where the word actually begins and ends, and that the reference transcription matches the spoken form. Where the machine got it wrong, you correct the boundary, split a segment, merge two, add a missing one, or remove one that contains no word.

Both collections run at 5.00 audit hours per task. Every judgment carries a written reason, so this is careful listening work rather than volume work.

## Who we are looking for

**Native or near-native command of Mandarin Chinese as spoken in mainland China.** You were either born and raised in a region where this language and variety is dominant, or you are fully fluent with five or more years of residence in such a region. This is a hard requirement.

**Strong English reading and writing.** All conventions, error codes, and audit rationale are written in English, so you need to read a detailed rulebook and write clear feedback in it.

**Disciplined rubric application.** The work rewards people who apply a written standard consistently across hundreds of judgments rather than relying on instinct. Prior transcription, subtitling, localization, linguistic annotation, or QA experience is a strong signal.

**Careful listening.** You can distinguish similar sounds, identify where one word ends and the next begins in continuous speech, and tell genuine acoustic ambiguity apart from a clear annotator error.

Helpful but not required: prior AI data annotation or model evaluation work, a phonetics or linguistics background, and experience with audio editing or waveform tools.

## What the work looks like

You listen, judge, and submit. Where you find an error you tag every applicable error code, not just the first one, and you explain in plain language why the task passed or failed. Critical errors are called out first so the reason for a failure is obvious to the reader.

You will complete a short calibration phase before production work begins.

Sonic Audit Specialist - Chinese

## The role

We are hiring native and near-native Korean speakers to **audit AI training data in their own language**, not to produce it.

You will work across two Amazon Sonic audit collections:

1. **Multilingual Transcription Audit.** You listen to a recording of human or agent speech, read the transcription an annotator produced, and judge whether it accurately captures what the speaker actually said. You apply a fixed set of error codes, assign a pass or fail verdict, and write a short rationale explaining the decision.

2. **Forced Alignment Audit.** You review audio that has already been segmented at the word level by an automated system. For each segment you verify that the start and end timestamps match where the word actually begins and ends, and that the reference transcription matches the spoken form. Where the machine got it wrong, you correct the boundary, split a segment, merge two, add a missing one, or remove one that contains no word.

Both collections run at 5.00 audit hours per task. Every judgment carries a written reason, so this is careful listening work rather than volume work.

## Who we are looking for

**Native or near-native command of Korean as spoken in South Korea.** You were either born and raised in a region where this language and variety is dominant, or you are fully fluent with five or more years of residence in such a region. This is a hard requirement.

**Strong English reading and writing.** All conventions, error codes, and audit rationale are written in English, so you need to read a detailed rulebook and write clear feedback in it.

**Disciplined rubric application.** The work rewards people who apply a written standard consistently across hundreds of judgments rather than relying on instinct. Prior transcription, subtitling, localization, linguistic annotation, or QA experience is a strong signal.

**Careful listening.** You can distinguish similar sounds, identify where one word ends and the next begins in continuous speech, and tell genuine acoustic ambiguity apart from a clear annotator error.

Helpful but not required: prior AI data annotation or model evaluation work, a phonetics or linguistics background, and experience with audio editing or waveform tools.

## What the work looks like

You listen, judge, and submit. Where you find an error you tag every applicable error code, not just the first one, and you explain in plain language why the task passed or failed. Critical errors are called out first so the reason for a failure is obvious to the reader.

You will complete a short calibration phase before production work begins.

Sonic Audit Specialist - Korean

## The role

We are hiring native and near-native Japanese speakers to **audit AI training data in their own language**, not to produce it.

You will work across two Amazon Sonic audit collections:

1. **Multilingual Transcription Audit.** You listen to a recording of human or agent speech, read the transcription an annotator produced, and judge whether it accurately captures what the speaker actually said. You apply a fixed set of error codes, assign a pass or fail verdict, and write a short rationale explaining the decision.

2. **Forced Alignment Audit.** You review audio that has already been segmented at the word level by an automated system. For each segment you verify that the start and end timestamps match where the word actually begins and ends, and that the reference transcription matches the spoken form. Where the machine got it wrong, you correct the boundary, split a segment, merge two, add a missing one, or remove one that contains no word.

Both collections run at 5.00 audit hours per task. Every judgment carries a written reason, so this is careful listening work rather than volume work.

## Who we are looking for

**Native or near-native command of Japanese as spoken in Japan.** You were either born and raised in a region where this language and variety is dominant, or you are fully fluent with five or more years of residence in such a region. This is a hard requirement.

**Strong English reading and writing.** All conventions, error codes, and audit rationale are written in English, so you need to read a detailed rulebook and write clear feedback in it.

**Disciplined rubric application.** The work rewards people who apply a written standard consistently across hundreds of judgments rather than relying on instinct. Prior transcription, subtitling, localization, linguistic annotation, or QA experience is a strong signal.

**Careful listening.** You can distinguish similar sounds, identify where one word ends and the next begins in continuous speech, and tell genuine acoustic ambiguity apart from a clear annotator error.

Helpful but not required: prior AI data annotation or model evaluation work, a phonetics or linguistics background, and experience with audio editing or waveform tools.

## What the work looks like

You listen, judge, and submit. Where you find an error you tag every applicable error code, not just the first one, and you explain in plain language why the task passed or failed. Critical errors are called out first so the reason for a failure is obvious to the reader.

You will complete a short calibration phase before production work begins.

Sonic Audit Specialist - Japanese

## The role

We are hiring native and near-native Brazilian Portuguese speakers to **audit AI training data in their own language**, not to produce it.

You will work across two Amazon Sonic audit collections:

1. **Multilingual Transcription Audit.** You listen to a recording of human or agent speech, read the transcription an annotator produced, and judge whether it accurately captures what the speaker actually said. You apply a fixed set of error codes, assign a pass or fail verdict, and write a short rationale explaining the decision.

2. **Forced Alignment Audit.** You review audio that has already been segmented at the word level by an automated system. For each segment you verify that the start and end timestamps match where the word actually begins and ends, and that the reference transcription matches the spoken form. Where the machine got it wrong, you correct the boundary, split a segment, merge two, add a missing one, or remove one that contains no word.

Both collections run at 5.00 audit hours per task. Every judgment carries a written reason, so this is careful listening work rather than volume work.

## Who we are looking for

**Native or near-native command of Portuguese as spoken in Brazil.** You were either born and raised in a region where this language and variety is dominant, or you are fully fluent with five or more years of residence in such a region. This is a hard requirement.

**Strong English reading and writing.** All conventions, error codes, and audit rationale are written in English, so you need to read a detailed rulebook and write clear feedback in it.

**Disciplined rubric application.** The work rewards people who apply a written standard consistently across hundreds of judgments rather than relying on instinct. Prior transcription, subtitling, localization, linguistic annotation, or QA experience is a strong signal.

**Careful listening.** You can distinguish similar sounds, identify where one word ends and the next begins in continuous speech, and tell genuine acoustic ambiguity apart from a clear annotator error.

Helpful but not required: prior AI data annotation or model evaluation work, a phonetics or linguistics background, and experience with audio editing or waveform tools.

## What the work looks like

You listen, judge, and submit. Where you find an error you tag every applicable error code, not just the first one, and you explain in plain language why the task passed or failed. Critical errors are called out first so the reason for a failure is obvious to the reader.

You will complete a short calibration phase before production work begins.

Sonic Audit Specialist - Portuguese

## The role

We are hiring native and near-native Italian speakers to **audit AI training data in their own language**, not to produce it.

You will work across two Amazon Sonic audit collections:

1. **Multilingual Transcription Audit.** You listen to a recording of human or agent speech, read the transcription an annotator produced, and judge whether it accurately captures what the speaker actually said. You apply a fixed set of error codes, assign a pass or fail verdict, and write a short rationale explaining the decision.

2. **Forced Alignment Audit.** You review audio that has already been segmented at the word level by an automated system. For each segment you verify that the start and end timestamps match where the word actually begins and ends, and that the reference transcription matches the spoken form. Where the machine got it wrong, you correct the boundary, split a segment, merge two, add a missing one, or remove one that contains no word.

Both collections run at 5.00 audit hours per task. Every judgment carries a written reason, so this is careful listening work rather than volume work.

## Who we are looking for

**Native or near-native command of Italian as spoken in Italy.** You were either born and raised in a region where this language and variety is dominant, or you are fully fluent with five or more years of residence in such a region. This is a hard requirement.

**Strong English reading and writing.** All conventions, error codes, and audit rationale are written in English, so you need to read a detailed rulebook and write clear feedback in it.

**Disciplined rubric application.** The work rewards people who apply a written standard consistently across hundreds of judgments rather than relying on instinct. Prior transcription, subtitling, localization, linguistic annotation, or QA experience is a strong signal.

**Careful listening.** You can distinguish similar sounds, identify where one word ends and the next begins in continuous speech, and tell genuine acoustic ambiguity apart from a clear annotator error.

Helpful but not required: prior AI data annotation or model evaluation work, a phonetics or linguistics background, and experience with audio editing or waveform tools.

## What the work looks like

You listen, judge, and submit. Where you find an error you tag every applicable error code, not just the first one, and you explain in plain language why the task passed or failed. Critical errors are called out first so the reason for a failure is obvious to the reader.

You will complete a short calibration phase before production work begins.

Sonic Audit Specialist - Italian

## The role

We are hiring native and near-native French speakers to **audit AI training data in their own language**, not to produce it.

You will work across two Amazon Sonic audit collections:

1. **Multilingual Transcription Audit.** You listen to a recording of human or agent speech, read the transcription an annotator produced, and judge whether it accurately captures what the speaker actually said. You apply a fixed set of error codes, assign a pass or fail verdict, and write a short rationale explaining the decision.

2. **Forced Alignment Audit.** You review audio that has already been segmented at the word level by an automated system. For each segment you verify that the start and end timestamps match where the word actually begins and ends, and that the reference transcription matches the spoken form. Where the machine got it wrong, you correct the boundary, split a segment, merge two, add a missing one, or remove one that contains no word.

Both collections run at 5.00 audit hours per task. Every judgment carries a written reason, so this is careful listening work rather than volume work.

## Who we are looking for

**Native or near-native command of French as spoken in France.** You were either born and raised in a region where this language and variety is dominant, or you are fully fluent with five or more years of residence in such a region. This is a hard requirement.

**Strong English reading and writing.** All conventions, error codes, and audit rationale are written in English, so you need to read a detailed rulebook and write clear feedback in it.

**Disciplined rubric application.** The work rewards people who apply a written standard consistently across hundreds of judgments rather than relying on instinct. Prior transcription, subtitling, localization, linguistic annotation, or QA experience is a strong signal.

**Careful listening.** You can distinguish similar sounds, identify where one word ends and the next begins in continuous speech, and tell genuine acoustic ambiguity apart from a clear annotator error.

Helpful but not required: prior AI data annotation or model evaluation work, a phonetics or linguistics background, and experience with audio editing or waveform tools.

## What the work looks like

You listen, judge, and submit. Where you find an error you tag every applicable error code, not just the first one, and you explain in plain language why the task passed or failed. Critical errors are called out first so the reason for a failure is obvious to the reader.

You will complete a short calibration phase before production work begins.

Sonic Audit Specialist - French

## The role

We are hiring native and near-native German speakers to **audit AI training data in their own language**, not to produce it.

You will work across two Amazon Sonic audit collections:

1. **Multilingual Transcription Audit.** You listen to a recording of human or agent speech, read the transcription an annotator produced, and judge whether it accurately captures what the speaker actually said. You apply a fixed set of error codes, assign a pass or fail verdict, and write a short rationale explaining the decision.

2. **Forced Alignment Audit.** You review audio that has already been segmented at the word level by an automated system. For each segment you verify that the start and end timestamps match where the word actually begins and ends, and that the reference transcription matches the spoken form. Where the machine got it wrong, you correct the boundary, split a segment, merge two, add a missing one, or remove one that contains no word.

Both collections run at 5.00 audit hours per task. Every judgment carries a written reason, so this is careful listening work rather than volume work.

## Who we are looking for

**Native or near-native command of German as spoken in Germany.** You were either born and raised in a region where this language and variety is dominant, or you are fully fluent with five or more years of residence in such a region. This is a hard requirement.

**Strong English reading and writing.** All conventions, error codes, and audit rationale are written in English, so you need to read a detailed rulebook and write clear feedback in it.

**Disciplined rubric application.** The work rewards people who apply a written standard consistently across hundreds of judgments rather than relying on instinct. Prior transcription, subtitling, localization, linguistic annotation, or QA experience is a strong signal.

**Careful listening.** You can distinguish similar sounds, identify where one word ends and the next begins in continuous speech, and tell genuine acoustic ambiguity apart from a clear annotator error.

Helpful but not required: prior AI data annotation or model evaluation work, a phonetics or linguistics background, and experience with audio editing or waveform tools.

## What the work looks like

You listen, judge, and submit. Where you find an error you tag every applicable error code, not just the first one, and you explain in plain language why the task passed or failed. Critical errors are called out first so the reason for a failure is obvious to the reader.

You will complete a short calibration phase before production work begins.

Sonic Audit Specialist - German

## The role

We are hiring native and near-native Spanish speakers to **audit AI training data in their own language**, not to produce it.

You will work across two Amazon Sonic audit collections:

1. **Multilingual Transcription Audit.** You listen to a recording of human or agent speech, read the transcription an annotator produced, and judge whether it accurately captures what the speaker actually said. You apply a fixed set of error codes, assign a pass or fail verdict, and write a short rationale explaining the decision.

2. **Forced Alignment Audit.** You review audio that has already been segmented at the word level by an automated system. For each segment you verify that the start and end timestamps match where the word actually begins and ends, and that the reference transcription matches the spoken form. Where the machine got it wrong, you correct the boundary, split a segment, merge two, add a missing one, or remove one that contains no word.

Both collections run at 5.00 audit hours per task. Every judgment carries a written reason, so this is careful listening work rather than volume work.

## Who we are looking for

**Native or near-native command of Spanish as spoken in Spain.** You were either born and raised in a region where this language and variety is dominant, or you are fully fluent with five or more years of residence in such a region. This is a hard requirement.

**Strong English reading and writing.** All conventions, error codes, and audit rationale are written in English, so you need to read a detailed rulebook and write clear feedback in it.

**Disciplined rubric application.** The work rewards people who apply a written standard consistently across hundreds of judgments rather than relying on instinct. Prior transcription, subtitling, localization, linguistic annotation, or QA experience is a strong signal.

**Careful listening.** You can distinguish similar sounds, identify where one word ends and the next begins in continuous speech, and tell genuine acoustic ambiguity apart from a clear annotator error.

Helpful but not required: prior AI data annotation or model evaluation work, a phonetics or linguistics background, and experience with audio editing or waveform tools.

## What the work looks like

You listen, judge, and submit. Where you find an error you tag every applicable error code, not just the first one, and you explain in plain language why the task passed or failed. Critical errors are called out first so the reason for a failure is obvious to the reader.

You will complete a short calibration phase before production work begins.

Sonic Audit Specialist - Spanish

Mercor is hiring experienced **private equity associates** to build a financial model from a synthetic deal document pack, then peer-review models built by two other contractors on the same transaction.

### What you'll do

- Build a working financial model of an assigned **consumer / leisure** transaction from a synthetic deal document pack
    
- Deliver an editable Excel workbook with live formulas, source references, and stated assumptions
    
- Peer-review two anonymized models from other contractors against a fixed set of criteria, with a short written rationale for each and a completed review scorecard
    

### Requirements

- **2–4 years total experience**: ~2 years investment banking (Consumer / Retail or generalist M&A — ideally Paris, London, or elsewhere in Europe) followed by 1–2 years in mid-market or large-cap private equity
    
- Currently or recently an **Associate at a mid-market or large-cap PE fund with a consumer / leisure / multi-site services focus** (fitness clubs, owned vs. franchise, multi-site roll-ups); a Paris office or French deal experience is preferred
    
- Able to build a financial model from source documents, without a template
    
- **English-speaking**
    

### Logistics

- **Effort:** ~20–25 hours over 1.5–2 weeks

Private Equity Associate – Consumer & Leisure (France/Europe)

Mercor is hiring experienced **private equity associates** to build a financial model from a synthetic deal document pack, then peer-review models built by two other contractors on the same transaction.

### What you'll do

- Build a working financial model of an assigned **real estate** transaction from a synthetic deal document pack
    
- Deliver an editable Excel workbook with live formulas, source references, and stated assumptions
    
- Peer-review two anonymized models from other contractors against a fixed set of criteria, with a short written rationale for each and a completed review scorecard
    

### Requirements

- **2–4 years total experience**: ~2 years investment banking (Real Estate or Industrials) followed by 1–2 years in private equity
    
- Currently or recently an **Associate at a real-estate-focused mid-market or upper-mid-market PE fund**
    
- Able to build a financial model from source documents, without a template
    
- Fluent **English**
    

### Logistics

- **Effort:** ~20–25 hours over 1.5–2 weeks

Private Equity Associate – Real Estate (US)

Mercor is hiring experienced **private equity associates** to build a financial model from a synthetic deal document pack, then peer-review models built by two other contractors on the same transaction.

### What you'll do

- Build a working financial model of an assigned **infrastructure** transaction from a synthetic deal document pack
    
- Deliver an editable Excel workbook with live formulas, source references, and stated assumptions
    
- Peer-review two anonymized models from other contractors against a fixed set of criteria, with a short written rationale for each and a completed review scorecard
    

### Requirements

- **2–4 years total experience**: ~2 years investment banking (infrastructure, power & utilities, or real estate / TMT M&A) followed by 1–2 years in infrastructure private equity
    
- Currently or recently an **Associate at an infrastructure-focused PE fund**, or the infrastructure arm of a multi-strategy manager
    
- Exposure to **digital infrastructure** — data centres, fibre, or towers
    
- Able to build a financial model from source documents, without a template
    
- Fluent **English**
    

### Logistics

- **Effort:** ~20–25 hours over 1.5–2 weeks

Private Equity Associate – Infrastructure (US/UK)

**Join a leading AI lab's cutting-edge GenAI team to be at the core of the AI revolution, where your expertise fuels the development of the most advanced AI models.**

## 1\. Overview

We are hiring experienced **mechanical engineers** to work directly with a leading AI lab's research and program management teams, improving how frontier AI models reason about real mechanical engineering work.

Your engineering judgment is the substance of this role. You will review the quality of mechanical engineering tasks and model outputs, write the instruction specs and reference solutions that define what "correct" looks like, and help build the evaluations that show whether the model is genuinely improving. We are looking for practicing engineers who have owned real designs, analyses, or systems in industry, not generalists.

This is a W-2 employment position with Cincinnatus LLC, with the opportunity to be placed at a leading AI lab as part of their extended workforce. The initial engagement is 2 to 3 months, with a strong possibility of extension based on performance and project needs. You will be provisioned with client-issued accounts and equipment, and will work inside the client's own tools alongside their research teams.

**Location:** Remote, within the United States.

## 2\. Key Responsibilities

- **Data QA and reviews:** Vet the quality of mechanical engineering tasks and model outputs, spotting incorrect assumptions, missing constraints, unit and tolerance errors, unsafe design choices, and answers that read well but would not pass a design review.
    
- **Instruction specs and reference solutions:** Write clear instruction specs, produce worked reference solutions to mechanical engineering problems, and define new tasks that reflect how engineering work is actually done in practice.
    
- **Benchmarks and domain depth:** Design challenging tasks and evaluation sets across areas such as mechanical design, thermal and fluid systems, structural and stress analysis, manufacturing, and controls, and help build domain-specific tools together with the research team.
    
- **Calibration:** Work with client researchers and specialists in adjacent fields to keep standards consistent, translating tacit engineering judgment into explicit, teachable criteria.
    

## 3\. Core Qualifications

- **Education:** Bachelor's degree or higher in Mechanical Engineering or a closely related engineering discipline.
    
- **Experience:** 5+ years of hands-on professional mechanical engineering experience in industry, for example product design and development, thermal and fluid systems, HVAC, structural or stress analysis, manufacturing engineering, automotive, aerospace, energy, robotics, or medical devices.
    
- **Domain depth:** Genuine ownership of real engineering work, for example designing components or systems, running FEA or CFD analyses, releasing drawings and specifications, or taking products through design reviews, testing, and production.
    
- **Tools and standards:** Working fluency with common engineering tools (for example SolidWorks, CATIA, NX, Creo, AutoCAD, ANSYS, Abaqus, or MATLAB) and familiarity with relevant codes and standards (for example ASME, ASTM, ISO, and GD&T per ASME Y14.5).
    
- **Seniority:** Clear progression toward senior or lead engineering responsibility. A Professional Engineer (PE) license is a plus but not required.
    
- **AI fluency:** Hands-on working use of large language models in your professional work, and the judgment to tell a well-reasoned answer from a plausible-sounding wrong one.
    
- **Availability:** Able to commit reliably to 40 hours per week for an initial engagement of 2 to 3 months, with a strong possibility of extension.
    
- Excellent written communication, and the ability to give precise, well-structured written feedback.
    

## About Cincinnatus LLC

Cincinnatus LLC is an enterprise staffing company that partners with leading technology companies to source and employ highly skilled professionals for contingent and contract-based opportunities. Cincinnatus serves as the employer of record for these engagements, providing W-2 employment, payroll, benefits, and compliance, while placing employees directly within client teams to work on high-impact initiatives.

Roles hired through Cincinnatus are not project-based or freelance engagements. They are structured, role-based positions that typically involve part-time or full-time commitments, close collaboration with a client's internal teams, and integration into standard enterprise workflows.

Cincinnatus is a legal entity separate from Mercor. While opportunities may be discovered through Mercor's platform, employment, onboarding, payroll, and benefits for these roles are administered by Cincinnatus LLC.

## Equal Employment Opportunity

Cincinnatus is proud to be an Equal Employment Opportunity employer. We do not discriminate based upon race, religion, color, national origin, sex (including pregnancy, childbirth, reproductive health decisions, or related medical conditions), sexual orientation, gender identity, gender expression, age, status as a protected veteran, status as an individual with a disability, genetic information, political views or activity, or any other legally protected characteristic.

Cincinnatus is committed to providing reasonable accommodations for qualified individuals with disabilities and disabled veterans throughout the job application process.

Mechanical Engineer

**What this is**
We're building a de-identified, real-world health dataset from Apple Health exports. Each contribution joins your on-device HealthKit data (activity, heart, sleep, and more) with clinical records (immunizations, labs, medications, vitals) from your linked health provider.

**Who qualifies**
We are exclusively looking for individuals who have visited at least one of the institutions listed below **and** have that institution's health records actively connected to Apple HealthKit. If your provider is not on this list, you do not qualify.

**Qualifying institutions:**

- Emory Healthcare
    
- Scripps Health
    
- Northwestern Medicine
    
- Cedars-Sinai Health System
    
- Penn Medicine
    
- NYU Langone Health
    
- Geisinger Health System
    
- Dartmouth Health
    
- Johns Hopkins Medicine
    
- UCSF Health
    
- Cleveland Clinic
    
- Jefferson Health
    
- VCU Health
    
- University of Chicago Medicine
    
- Doctors Center Hospital Orlando Health
    
- Mass General Brigham
    
- Norton Healthcare
    
- Mount Sinai Health System
    
- Henry Ford Health System
    
- Yale New Haven Health
    
- Intermountain Healthcare
    
- Tampa General Hospital
    
- The University of Kansas Health System
    
- University of Utah Health
    
- Michigan Medicine
    
- UC Irvine Health
    
- Nebraska Medicine
    
- Vanderbilt University Medical Center
    
- UC San Diego Health
    
- Musculoskeletal Institute Of LA
    
- UVA Health System
    
- University of Iowa Health Care
    
- Memorial Hermann Health System
    
- Mount Sinai Medical Center
    
- Sentara Healthcare
    
- Froedtert & the Medical College of Wisconsin
    
- Stanford Medicine Obstetrics & Reproductive Endocrinology
    
- Inova Health System
    
- Ochsner Health
    
- UC Davis Health
    
- Scripps Ranch Medical Clinic
    
- The Ohio State University Wexner Medical Center
    
- Buffalo Neuromusculoskeletal Medicine PLLC
    
- The George Washington University Medical Faculty Associates
    
- Houston Methodist
    
- Stanford Health Care
    
- Baylor Scott & White Health
    
- UnityPoint Health
    
- Sutter Health
    
- Aurora Health Care
    
- RWJBarnabas Health
    
- Trinity Health IHA Medical Group
    
- Valley Health System
    
- CHI St. Joseph Health
    
- M Health Fairview
    
- Hattiesburg Clinic
    
- The Christ Hospital Health Network
    
- Bon Secours Health System
    
- Kaiser Permanente
    
- Methodist Le Bonheur Healthcare
    

**What you'll do**

1. **Fully populate your HealthKit profile**: add data across every category that applies to you. Steps-only profiles won't qualify.
    
2. **Link your health record (EHR)** from at least one of the qualifying institutions above via Apple's Health Records network. Must include immunizations, lab results, medications, vitals, **and clinical notes**.
    
3. **Consent to share four demographics**: age, gender, approximate region, and activity level.
    

**To qualify, your submission must:**

- Be a genuine export from your own Health app (not synthetic or demo)
    
- Include records from at least one institution on the qualifying list above, connected to Apple HealthKit
    
- Show a complete profile with continuous history, not a single day of data
    
- Include a linked EHR with all four required clinical categories
    
- **Include the Clinical Notes section in your health record — this is a hard requirement. Submissions without clinical notes cannot be accepted or compensated and will be returned.**
    
- Have all four demographic attributes consented — **no consent, no submission**
    

**Your privacy**
Identity and raw health information never reach the dataset. Every export is de-identified (HIPAA Safe-Harbor–style) before analysis: identifiers stripped, dates and locations generalized, notes scrubbed. You get a random ID; the link to your identity is held separately and never travels with the data. Withdraw consent anytime to be removed.

**Commitment**
One export per participant, with an optional re-export later to refresh your data.

Apple Watch Users: Opportunity for You

Get paid to support Wellness Research (US Only)

Fitness Data Contributor

iPhone Users: Earn Money for Sharing Your Fitness Data (US Only)

**Join a leading AI lab's cutting-edge GenAI team to be at the core of the AI revolution, where your expertise fuels the development of the most advanced Large Language Models.**

# 1\. Overview

Join a leading AI lab's cutting-edge GenAI team and help build foundational AI models from the ground up. We're seeking MLOps Engineers with hands-on experience in large language model infrastructure across any of four areas: GPU kernel programming, performance profiling and trace analysis, debugging accelerated and distributed workloads, and high-throughput inference serving. This role involves AI model training and evaluation work, including writing and assessing MLOps and ML systems tasks and solutions to generate high-quality training data for frontier AI systems.

This is a W-2 employment position with Cincinnatus LLC, with the opportunity to be placed at a leading AI Lab as part of their extended workforce. **This is a 40-hour full-time engagement, with no conflicts/no other engagements.**

# 2\. Key Responsibilities

- Design challenging, domain-relevant tasks across four areas, GPU kernels, performance profiling, debugging, and inference serving, and write accurate, well-structured solutions to them.
    
- Guide research and engineering teams to close knowledge gaps and improve AI model performance on ML systems, training infrastructure, and framework-level topics.
    
- Evaluate MLOps and ML systems tasks and solutions, and provide clear, written technical feedback that stands up to reviewer scrutiny.
    
- Develop guidelines and detailed rubrics or evaluation frameworks covering kernel-level optimization, profiler output interpretation, distributed systems reasoning, and serving throughput and latency trade-offs.
    
- Collaborate with other subject matter experts to keep training data consistent and accurate.
    

# 3\. Core Qualifications

- 2+ years of hands-on professional experience in ML systems, ML infrastructure, model serving, or GPU and accelerator performance engineering. This is a hands-on systems role rather than an applied modelling or data science one.
    
- Practical experience in at least one of the following, with more than one a strong plus: writing or optimizing custom GPU kernels (CUDA, Triton, Pallas); performance profiling and trace analysis (Kineto, torch.profiler, Nsight, XLA or JAX profiler); debugging distributed or accelerator-bound workloads; serving large language models at scale (vLLM, SGLang, TensorRT-LLM, Ray Serve, KV cache, paged attention, continuous batching).
    
- Working production experience with JAX and/or PyTorch. Framework-level depth is a strong plus: custom operators, distributed training (FSDP, DDP, DeepSpeed, Megatron), or compiler and graph-level work.
    
- Familiarity with modern accelerators such as A100, H100, B200 or TPU, and the ability to reason about throughput, latency and memory trade-offs.
    
- Demonstrable career progression.
    
- Ability to engage reliably for at least 40 hours/week during weekdays.
    
- Strong written communication skills and the ability to explain complex technical decisions clearly.
    

**About Cincinnatus LLC:**
Cincinnatus LLC is an enterprise staffing company that partners with leading technology companies to source and employ highly skilled professionals for contingent and contract-based opportunities. Cincinnatus serves as the employer of record for these engagements, providing W-2 employment, payroll, benefits, and compliance, while placing employees directly within client teams to work on high-impact initiatives.

**Equal Employment Opportunity:**
Cincinnatus is proud to be an Equal Employment Opportunity employer. We do not discriminate based upon race, religion, color, national origin, sex (including pregnancy, childbirth, reproductive health decisions, or related medical conditions), sexual orientation, gender identity, gender expression, age, status as a protected veteran, status as an individual with a disability, genetic information, political views or activity, or any other legally protected characteristic.

MLOps Engineer, LLM Systems (Serving, GPU Kernels, Profiling)

The research teams behind the best-known AI models come to Mercor for careful human judgment they can't generate on their own. Your standards and attention become the rubric that frontier AI is graded against.

This is a standing listing for generalists who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match generalists in the network to projects like:

- Writing prompts and reference answers on everyday questions that have a right answer, with the context a good response has to account for
    
- Grading model output on general reasoning: whether the answer is correct, whether it follows the instruction actually given, whether it is complete without padding
    
- Flagging where a model is confidently wrong, contradicts itself earlier in the same answer, or answers a question that was not asked
    

Most contracts in this field have paid $40 to $70 an hour, set per project by scope and depth.

**What we're looking for**

- Careful reading and sound judgment, from a professional or academic background in any field
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

**Who this is for**

This pool is for people whose work does not fit neatly into one of our specialist networks, or who work across several of them. If you looked through our other listings and none of them quite described you, this is the one.

No particular credential is required. What matters is that you read carefully and can say precisely why an answer is wrong.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the generalist pool.

When a project needs general reasoning work, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Generalist Expert

The research teams behind the best-known AI models come to Mercor for data judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for analysts and data practitioners who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match data practitioners in the network to projects like:

- Writing an analysis problem from work you actually shipped, with the data, the question, and the constraint, plus the reference answer a model's attempt gets graded against
    
- Grading model-written analysis and pipelines: whether the query answers the question that was asked, whether the join fans out, whether the metric means what the model says it means
    
- Flagging where a model produces a technically valid result that would mislead a decision-maker, given selection effects, leakage, or a metric that does not track the outcome
    

**What we're looking for**

- Professional experience in analytics, data engineering, statistics, or machine learning
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

**What this covers**

This pool spans data analysis, including:

- Business intelligence
    
- Data engineering
    
- Statistics
    
- Machine learning
    

Where you trained matters less than whether you have shipped analyses someone made a decision on.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the data analysis pool.

When a project needs your specialty, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Data Analysis Expert

The research teams behind the best-known AI models come to Mercor for teaching judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for educators who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match educators in the network to projects like:

- Writing a teaching problem from your own classroom or course, with the level, the common misconception, and the constraint, plus the reference answer a model's attempt gets graded against
    
- Grading model-written explanations and lesson material: whether the explanation meets the learner where they are, whether the worked example is correct, whether the assessment measures what it claims to
    
- Flagging where a model's explanation is accurate and would still confuse a student, given prior knowledge, reading level, or the misconception it fails to address
    

**What we're looking for**

- Professional experience teaching, tutoring, or designing instruction
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

**What this covers**

This pool spans education and training, including:

- K-12 teaching
    
- Higher education and academia
    
- Tutoring and test prep
    
- Curriculum and instructional design
    
- ESL and language instruction
    
- Education technology
    

Where you trained matters less than whether you have taught the material and seen where learners get stuck.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the education and training pool.

When a project needs your subject or level, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Education & Training Expert

The research teams behind the best-known AI models come to Mercor for judgment about how office and productivity software is actually used. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for people who work in these tools all day and who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match people in this network to projects like:

- Writing a realistic task from your own work, a document, a model, or a deck, with the starting material and the constraint, plus the reference answer a model's attempt gets graded against
- Grading model-generated documents, spreadsheets, and slides: whether the formula is correct and robust, whether the formatting survives editing, whether the deck actually makes its argument
- Flagging where a model's output opens cleanly and would break in real use, given file conversion, version compatibility, or what happens once someone else edits it

Listings in this field on Mercor have posted at $60 to $70 an hour, set per project by scope and depth.

**What we're looking for**

- Professional experience working at a high level in office and productivity tools
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**What this covers**

This pool spans office and productivity software, including:

- Documents and word processing
- Spreadsheets and modeling
- Presentations and slides
- PDF and document conversion
- Email and collaboration tools
- Office scripting and add-ins

Depth in one tool matters more than passing familiarity with all of them.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the office and productivity software pool.

When a project needs your tool or specialty, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Office & Productivity Software Expert

The research teams behind the best-known AI models come to Mercor for financial judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for finance professionals who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match finance professionals in the network to projects like:

- Writing a financial problem from work you have done, with the filings, the assumptions, and the constraints, plus the reference answer a model's attempt gets graded against
- Grading model-built financial work: whether the model ties, whether the accounting treatment is right, whether the valuation assumptions survive a sensitivity check
- Flagging where a model's output looks clean on the page and would not pass review, given disclosure requirements, audit standards, or what the numbers actually support

Rates vary by specialty within this field. Listings across it have posted from $60 to $180 an hour, set per project by scope and depth.

**What we're looking for**

- Professional experience in a finance or accounting discipline
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**What this covers**

This pool spans the finance field, including:

- Accounting and auditing
- Corporate finance; investment banking
- Asset management; wealth management
- Insurance and actuarial science

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the finance pool.

When a project needs your specialty, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

The research teams behind the best-known AI models come to Mercor for humanities judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for humanities scholars who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match scholars in the network to projects like:

- Writing a question from your own field, with the sources, the context, and the interpretive problem, plus the reference answer a model's attempt gets graded against
- Grading model-written interpretation and argument: whether the sources say what the model claims, whether the reasoning holds, whether the historical or cultural context is right
- Flagging where a model produces fluent, confident prose that misreads the text, flattens a live scholarly debate, or invents a citation

Listings in this field on Mercor have posted at $45 to $60 an hour, set per project by scope and depth.

**What we're looking for**

- A graduate degree in a humanities field, or equivalent scholarly or professional experience
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**What this covers**

This pool spans the humanities, including:

- History; classics
- Philosophy; ethics and logic
- Literature
- Religious studies; cultural studies

Where you trained matters less than whether you can tell a strong reading from a merely plausible one, and say precisely where the weaker one fails.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the humanities pool.

When a project needs your field, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Humanities Expert

The research teams behind the best-known AI models come to Mercor for design and creative judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for designers and creative professionals who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match creative professionals in the network to projects like:

- Writing a brief from work you have delivered, with the audience, the constraints, and the standard the result had to meet, plus the reference answer a model's attempt gets graded against
- Grading model-generated design, writing, and imagery on craft: whether the hierarchy works, whether the type and colour choices are deliberate, whether the prose earns its length
- Flagging where a model's output is technically competent and would not ship, given brand, accessibility, licensing, or what the medium actually allows

Creative listings on Mercor have posted at $60 to $80 an hour, set per project by scope and depth.

**What we're looking for**

- Professional experience in design, writing, or a creative production discipline
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**What this covers**

This pool spans arts and design, including:

- Graphic and UX/UI design
- Visual arts, including painting and sculpture
- Performing arts, including music and theatre
- Film and media production; photography
- Fashion and textile design
- Writing and editing

A portfolio matters more than a credential here. What we most need is your ability to say why something works.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the arts and design pool.

When a project needs your discipline, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Arts & Design Expert

The research teams behind the best-known AI models come to Mercor for legal judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for legal professionals who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match legal professionals in the network to projects like:

- Writing a legal problem from a matter you worked on, with the facts, the forum, and the governing law, plus the reference answer a model's attempt gets graded against
- Grading model-drafted legal work: whether the provision does what it claims, whether the privilege call is right, whether the citation supports the proposition it is offered for
- Flagging where a model states the law correctly and would still lose, given the forum, the standard of review, or facts that do not carry the argument

Rates vary by role within this field. Listings across it have posted from $60 to $150 an hour, set per project by scope and depth.

**What we're looking for**

- Professional experience in legal practice, compliance, regulatory work, or policy
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**What this covers**

This pool spans the legal field, including:

- Legal practice; criminal justice and litigation
- Corporate law and M&A; intellectual property
- Compliance and risk management; regulatory affairs
- Policy
- Paralegals and legal assistants

This pool is broader than practising attorneys. Paralegals, compliance specialists, and policy professionals belong here too.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the legal pool.

When a project needs your area of practice, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Legal Expert

The research teams behind the best-known AI models come to Mercor for engineering judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for engineers who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match engineers in the network to projects like:

- Writing a design problem from work you have done, with the loads, the constraints, and the tolerances, plus the reference answer a model's attempt gets graded against
- Grading model-written analysis: whether the governing failure mode is the one considered, whether the assumptions hold in the regime claimed, whether the safety factor is applied to the right quantity
- Flagging where a model's design satisfies the calculation and fails in practice, given manufacturability, installation, corrosion, or what the standard actually requires

Engineering projects on Mercor have posted at $60 to $125 an hour, set per project by scope and depth.

**What we're looking for**

- A degree in an engineering discipline, with professional design or analysis experience
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**What this covers**

This pool spans the engineering disciplines, including:

- Structural and civil engineering; environmental engineering
- Mechanical, aerospace, and industrial engineering; robotics
- Electrical engineering
- Chemical engineering; materials engineering

Where you trained matters less than whether you have had a design of yours built, and learned what it did once it was.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the engineering pool.

When a project needs your discipline, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Engineering Expert

The research teams behind the best-known AI models come to Mercor for software engineering judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for engineers who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match engineers in the network to projects like:

- Writing an engineering problem from a system you have shipped, with the requirements, the constraints, and the failure modes, plus the reference answer a model's attempt gets graded against
- Grading model-written code and design: whether it handles the edge cases, whether the complexity is justified, whether the test actually exercises the behaviour it claims to
- Flagging where a model's code passes review and would fail in production, given concurrency, scale, error handling, or what it costs to operate

Software engineering listings on Mercor have posted at $70 to $150 an hour, set per project by scope and depth.

**What we're looking for**

- Professional experience building and shipping software
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**What this covers**

This pool spans software engineering, including:

- Frontend (React, TypeScript, Next.js); backend (Java, Go, Node.js); full-stack
- Mobile (Android, iOS); embedded and IoT
- DevOps and platform engineering; systems and performance engineering
- QA and automation; security engineering

Where you trained matters less than whether you have owned code in production and dealt with what it did there.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the software engineering pool.

When a project needs your stack or specialty, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Software Engineering Expert

The research teams behind the best-known AI models come to Mercor for consumer and lifestyle judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for specialists in consumer categories who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match specialists in the network to projects like:

- Writing a realistic consumer question from your own area, with the context and constraints a good answer has to respect, plus the reference answer a model's attempt gets graded against
- Grading model recommendations on whether they are current, practical, and safe: whether the product actually exists, whether the advice suits the budget and the setting, whether the cultural detail is right
- Flagging where a model's answer sounds confident and would embarrass someone who acted on it, given price, availability, seasonality, or local norms

Listings in this field on Mercor have posted at $50 to $75 an hour, set per project by scope and depth.

**What we're looking for**

- Professional or deep practical experience in a consumer category
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**What this covers**

This pool spans consumer and lifestyle categories, including:

- Sports and athletics; fitness and wellness
- Film, TV and pop culture; gaming and esports
- Food and drink; travel and geography
- Fashion and personal style; shopping and retail
- Consumer electronics and smart home; home and family life

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the consumer and lifestyle pool.

When a project needs your category, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Consumer & Lifestyle Expert

The research teams behind the best-known AI models come to Mercor for medical judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for clinicians and health professionals who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match health professionals in the network to projects like:

- Writing a case from your own practice, with the presentation, the workup, and the constraints, plus the reference answer a model's attempt gets graded against
    
- Grading model output on clinical reasoning: whether the differential is ordered sensibly, whether the guideline actually applies to this patient, whether the plan is safe to carry out
    
- Flagging where a recommendation is defensible in a textbook and wrong in front of the patient, given comorbidities, access, or what the care setting can deliver
    

Rates vary widely by profession within this field. Listings across it have posted from $60 to $180 an hour, set per project by scope and depth.

**What we're looking for**

- A clinical credential and professional experience in a health profession
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

**What this covers**

This pool spans the medical field, including:

- Internal medicine, surgical specialties, and diagnostic medicine
    
- Mental health, including psychiatry and clinical psychology
    
- Nursing; dentistry; pharmacy
    
- Medical technology and laboratory science
    
- Public health and epidemiology
    
- Healthcare administration; medical writing
    

This pool is broader than physicians. If you carry clinical judgment in any health profession, it belongs here.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your credentials and background and add you to the medicine pool.

When a project needs your profession or specialty, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Medical Expert

The research teams behind the best-known AI models come to Mercor for scientific judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for researchers who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match researchers in the network to projects like:

- Writing a research problem from your own field, with the system, the data, and the assumptions, plus the reference answer a model's attempt gets graded against
    
- Grading model-written analysis and interpretation: whether the method suits the question, whether the statistics support the claim, whether the conclusion follows from the evidence
    
- Flagging where a model produces a result that is internally consistent and would not survive peer review, given the design, the sample, or what the measurement can actually show
    

Projects across these fields have posted at $60 to $120 an hour, set per project by scope and depth.

**What we're looking for**

- A graduate degree in a life, physical, or social science, with research experience
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

**What this covers**

This pool spans the sciences, including:

- Biology, biotechnology, chemistry, and materials science
    
- Physics and astronomy; mathematics
    
- Earth and environmental science; agriculture
    
- Psychology; sociology and anthropology
    
- Economics; political science and policy research
    

Where you trained matters less than whether you designed the studies and interpreted the data yourself, recently enough to explain the calls you made.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the science pool.

When a project needs your field, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Life, Physical & Social Science Expert

The research teams behind the best-known AI models come to Mercor for business operations judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for operators who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match operators in the network to projects like:

- Writing an operating problem from work you have run, with the process, the constraint, and the numbers, plus the reference answer a model's attempt gets graded against
- Grading model-written operational analysis: whether the process change would survive contact with the people who run it, whether the forecast is grounded, whether the metric actually tracks the outcome
- Flagging where a model's recommendation is defensible on a slide and would stall in practice, given headcount, systems, vendor terms, or who has to approve it

Comparable business operations projects have posted at $90 to $120 an hour, set per project by scope and depth.

**What we're looking for**

- Professional experience running or analyzing a business function
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**What this covers**

This pool spans the business operations field, including:

- Sales and business development; customer success and account management
- Strategy, process improvement, and operations analysis
- Supply chain and logistics; procurement
- Project and people management; human resources and administration
- Marketing; survey and market research; customer support operations

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the business operations pool.

When a project needs your function, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Business Operations Expert

The research teams behind the best-known AI models come to Mercor for language and audio judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for linguists, translators, and audio professionals who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match people in this network to projects like:

- Writing prompts and reference answers in your language, with the register, idiom, and cultural context a model has to get right for the answer to read as native
- Grading model output in your language: whether a translation preserves meaning and tone, whether the grammar is idiomatic rather than merely correct, whether the register suits the speaker and the setting
- Judging voice, music, and sound work on whether it is clean, natural, and usable, and saying precisely what is wrong when it is not

Listings in this field on Mercor have posted at $35 to $50 an hour, set per project by scope, language, and depth.

**What we're looking for**

- Native or professional fluency in at least one language, or professional experience in translation, voice, music, or audio production
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**What this covers**

This pool spans the language and audio field, including:

- Languages: Spanish, Arabic, English, Hindi, French, Portuguese, Italian, Chinese, Russian, Japanese, Korean, Polish, Tamil, Telugu, Kannada, Malayalam, Marathi, Assamese, Punjabi, Gujarati, Thai, Dutch, and Finnish
- Translation and interpretation
- Voice acting
- Music, audio engineering, and sound design

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the language and audio pool.

When a project needs your language or your specialty, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Language & Audio Expert

The research teams behind the best-known AI models come to Mercor for electrical engineering judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for electrical engineers who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match electrical engineers in the network to projects like:

- Writing a circuit or systems problem from work you have done, with the specification, the constraints, and the operating conditions, plus the reference answer a model's attempt gets graded against
    
- Grading model-written design and analysis: whether the topology suits the requirement, whether the noise and stability margins hold, whether the timing closes
    
- Flagging where a model's design simulates cleanly and fails on a real board, given parasitics, tolerance stack-up, or interference
    

Past rates have run $70 to $120 an hour, set per project by scope and depth.

**What we're looking for**

- A degree in electrical engineering, with professional design, test, or systems experience
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

Where you trained matters less than whether you have debugged your own design on the bench and found out what the simulation missed.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the electrical engineering pool.

When a project needs electrical engineering expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Electrical Engineering Expert

The research teams behind the best-known AI models come to Mercor for mechanical engineering judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for mechanical engineers who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match mechanical engineers in the network to projects like:

- Writing a design problem from work you have done, with the loads, the constraints, and the tolerances, plus the reference answer a model's attempt gets graded against
    
- Grading model-written analysis: whether the free body diagram is right, whether the failure mode considered is the governing one, whether the safety factor is applied to the right quantity
    
- Flagging where a model's design satisfies the calculation and fails in practice, given manufacturability, assembly, wear, or thermal expansion
    

Past rates have run $70 to $125 an hour, set per project by scope and depth.

**What we're looking for**

- A degree in mechanical engineering, with professional design or analysis experience
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

Where you trained matters less than whether you have had a design of yours built, and learned what it did once it was.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the mechanical engineering pool.

When a project needs mechanical engineering expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Mechanical Engineering Expert

The research teams behind the best-known AI models come to Mercor for mathematical judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for mathematicians who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match mathematicians in the network to projects like:

- Writing a problem at research or competition level, with the statement and the intended solution, plus the reference answer a model's attempt gets graded against
    
- Grading model-written proofs: whether each step follows, whether the edge cases are handled, whether an appeal to a known theorem actually satisfies its hypotheses
    
- Flagging where a model reaches a correct answer through an argument that does not prove it
    

Most contracts in this field have paid $40 to $75 an hour, set per project by scope and depth.

**What we're looking for**

- A graduate degree in mathematics or a related field, or an equivalent research or competition background
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

Where you trained matters less than whether you can tell a proof from a convincing sketch, and say precisely where the gap is.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the mathematics pool.

When a project needs mathematics expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Mathematics Expert

The research teams behind the best-known AI models come to Mercor for physics judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for physicists who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match physicists in the network to projects like:

- Writing a problem from your own area, with the system, the assumptions, and the regime, plus the reference answer a model's attempt gets graded against
    
- Grading model-written derivations and numerical work: whether the approximation is valid in the regime claimed, whether the limiting cases check out, whether the units and scaling survive scrutiny
    
- Flagging where a model reaches the right final expression through reasoning that would not hold if the assumptions changed
    

Past rates have run $70 to $100 an hour, set per project by scope and depth.

**What we're looking for**

- A graduate degree in physics or a related field, with research experience
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

Where you trained matters less than whether you did the derivations and the analysis yourself, recently enough to explain the calls you made.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the physics pool.

When a project needs physics expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Physics Expert

The research teams behind the best-known AI models come to Mercor for materials science judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for materials scientists who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match materials scientists in the network to projects like:

- Writing a materials problem from work you have done, with the system, the processing history, and the property target, plus the reference answer a model's attempt gets graded against
    
- Grading model-written structure-property reasoning: whether the microstructure explains the property, whether the characterization supports the claim, whether the processing window is realistic
    
- Flagging where a model's proposed material meets the spec on paper and fails in service, given fatigue, corrosion, thermal cycling, or what can actually be manufactured
    

Past rates have run $60 to $120 an hour, set per project by scope and depth.

**What we're looking for**

- A graduate degree in materials science or a related field, with research or industrial experience
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

Where you trained matters less than whether you processed and characterized the material yourself, recently enough to explain the calls you made.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the materials science pool.

When a project needs materials science expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Materials Science Expert

The research teams behind the best-known AI models come to Mercor for chemistry judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for chemists who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match chemists in the network to projects like:

- Writing a synthesis or characterization problem from your own work, with the substrate, the conditions, and the analytical data, plus the reference answer a model's attempt gets graded against
    
- Grading model-proposed routes and mechanisms: whether the disconnection is sensible, whether the selectivity claim is supported, whether the spectra match the structure assigned
    
- Flagging where a model proposes something that works on paper and fails in the flask, given side reactivity, workup losses, or what the reagent tolerates
    

Past rates have run $65 to $105 an hour, set per project by scope and depth.

**What we're looking for**

- A graduate degree in chemistry or a related field, with laboratory or computational research experience
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

Where you trained matters less than whether you ran the reactions and interpreted the data yourself, recently enough to explain the calls you made.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the chemistry pool.

When a project needs chemistry expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Chemistry Expert

The research teams behind the best-known AI models come to Mercor for biotechnology judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for biotech scientists who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match biotech scientists in the network to projects like:

- Writing a development problem from a program you worked on, with the modality, the process step, and the specification it had to meet, plus the reference answer a model's attempt gets graded against
- Grading model-written process and assay development work: whether the scale-up assumptions hold, whether the potency assay measures what it claims, whether the release criteria are justified
- Flagging where a model's proposal is scientifically sound and would still fail a developability assessment, given manufacturability, stability, or what a regulator would accept

Most contracts in this field have paid $110 to $125 an hour, set per project by scope and depth.

**What we're looking for**

- Professional or research experience in biotechnology, bioprocess, or therapeutic development
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

Where you trained matters less than whether you took a program through real development decisions and can explain why each one went the way it did.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the biotechnology pool.

When a project needs biotechnology expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Biotechnology Expert

The research teams behind the best-known AI models come to Mercor for biology judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for biologists who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match biologists in the network to projects like:

- Writing a research problem from your own work, with the system, the data, and the controls, plus the reference answer a model's attempt gets graded against
    
- Grading model-written experimental designs and interpretations: whether the controls rule out what they claim to, whether the statistics suit the design, whether the conclusion follows from the figure
    
- Flagging where a model gives a textbook-correct answer that would not survive at the bench, given the organism, the limits of the assay, or what the reagents can actually show
    

Past rates have run $70 to $105 an hour, set per project by scope and depth.

**What we're looking for**

- A graduate degree in biology or a related life science, with research experience
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

Where you trained matters less than whether you designed and ran the experiments yourself, recently enough to explain the calls you made.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the biology pool.

When a project needs biology expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Biology Expert

The research teams behind the best-known AI models come to Mercor for machine learning judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for machine learning practitioners who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match machine learning practitioners in the network to projects like:

- Writing a modeling problem from a system you actually shipped, with the data, the objective, and the constraint, plus the reference answer a model's attempt gets graded against
    
- Grading model-written training and evaluation code: whether the loss matches the objective, whether the eval set leaks, whether the ablation supports the conclusion drawn from it
    
- Flagging where a model proposes an approach that would train cleanly and still fail in production, given distribution shift, label noise, or a metric that rewards the wrong behaviour
    

Comparable machine learning projects have posted at $70 to $120 an hour, set per project by scope and depth.

**What we're looking for**

- Professional experience training, evaluating, or deploying machine learning systems
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

Where you trained matters less than whether you have put models in front of real users and dealt with what came back.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the machine learning pool.

When a project needs machine learning expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Machine Learning Expert

The research teams behind the best-known AI models come to Mercor for investment management judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for asset and wealth managers who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match asset and wealth managers in the network to projects like:

- Writing an investment case or a client scenario from your own mandate or book, with the thesis, the constraints, and the risk budget, plus the reference answer a model's attempt gets graded against
- Grading model-built portfolio construction, allocation, and tax strategy: whether the factor exposures are intended, whether the benchmark is the right one, whether the withdrawal sequencing holds, whether the attribution actually explains the return
- Flagging where a model's recommendation is defensible on paper and would breach a real mandate or fail a real client, given liquidity, concentration limits, drawdown tolerance, or time horizon

Comparable investment management projects have posted at $100 to $130 an hour, set per project by scope and depth.

**What we're looking for**

- Professional experience in portfolio management, buy-side research, multi-asset allocation, financial planning, or private client advisory
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**Where people in this pool have worked**

The most common backgrounds, though the pool is broader than this:

- Long-only asset managers and mutual fund complexes
- Hedge funds and multi-strategy platforms
- Pension, endowment, and sovereign allocators
- Private banks and wirehouse wealth management
- Registered investment advisors and multi-family offices
- Trust, estate, and fiduciary practices

The firm name matters less than whether you made the calls yourself, recently enough to explain the tradeoffs behind them.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the asset and wealth management pool.

When a project needs asset or wealth management expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Asset & Wealth Management Expert

The research teams behind the best-known AI models come to Mercor for management consulting judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for consultants who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match consultants in the network to projects like:

- Writing a case prompt from an engagement you ran, with the client situation, the data, and the constraint, plus the reference answer a model's attempt gets graded against
    
- Grading model-built market sizings and profitability analyses: whether the segmentation is sensible, whether the driver tree holds, whether the recommendation follows from the numbers
    
- Flagging where a model produces a clean-looking answer that would not survive a partner review or a client steering committee
    

Past rates have run $90 to $120 an hour, set per project by scope and depth.

**What we're looking for**

- Professional experience in strategy, operations, or transformation consulting
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

**Where people in this pool have worked**

The most common backgrounds, though the pool is broader than this:

- MBB and the strategy arms of the Big Four
    
- Boutique and tier-two strategy firms
    
- Functional specialists: pricing, operations, org design, post-merger integration
    
- In-house strategy, corporate development, or internal consulting at a large company
    

The firm name matters less than whether you owned the analysis and the client-facing recommendation, not just the slide production.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the management consulting pool.

When a project needs management consulting expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

The research teams behind the best-known AI models come to Mercor for legal judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for lawyers who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match lawyers in the network to projects like:

- Drafting a motion or a brief from a fact pattern, plus the reference answer a model's attempt gets graded against
    
- Grading model-drafted contract and discovery work: whether the provision does what it claims, whether the privilege call is right, whether the citation supports the proposition it is offered for
    
- Flagging where a model states the law correctly and would still lose, given the forum, the standard of review, or facts that do not carry the argument
    

Projects in this field have posted at $100 to $150 an hour, set per project by scope and depth.

**What we're looking for**

- A J.D., LL.B., or LL.M. and a current license to practice
    
- Professional experience in litigation, transactional work, or regulatory practice
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

**Where people in this pool have worked**

The most common backgrounds, though the pool is broader than this:

- Am Law firms and litigation or transactional boutiques
    
- Federal and state judicial clerkships
    
- In-house counsel and compliance at large companies
    
- Federal and state enforcement, regulatory, and prosecutorial offices
    

The firm name matters less than whether you did the drafting and the analysis yourself, recently enough to explain why you came out where you did.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your license and background and add you to the legal pool.

When a project needs your area of practice, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Legal Expert (J.D./LL.B.)

The research teams behind the best-known AI models come to Mercor for clinical judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for physicians who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match physicians in the network to projects like:

- Writing a clinical vignette from your own practice, with the presentation, labs, and imaging, plus the attending-level reference answer a model gets graded against
    
- Grading model output on diagnostic reasoning: whether the differential is ordered sensibly, whether it anchors, whether the proposed workup is guideline-concordant
    
- Flagging where a recommendation is defensible in a textbook but wrong for the patient in front of you, given comorbidities, contraindications, or what the patient can access
    

**What we're looking for**

- An M.D. or D.O. with board certification and a current, active, unrestricted license
    
- Active clinical practice, or recent enough that you can still explain the reasoning behind a call
    
- Clear written communication, since most of the work involves explaining your reasoning
    
- Comfort with ambiguous tasks, and willingness to flag unclear instructions
    

Final-year residents and fellows are welcome. Where you trained matters less than whether you carry current clinical judgment.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your license and background and add you to the physician pool.

When a project needs your specialty, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Medical Expert (M.D./D.O.)

The research teams behind the best-known AI models come to Mercor for investment banking judgment they can't generate on their own. Your standards and expertise become the rubric that frontier AI is graded against.

This is a standing listing for investment bankers who want AI training and evaluation work, not a specific job opening. It stays open whether or not a matching project is running this week.

New projects start on short notice, so instead of posting and pulling a listing each time, we keep one open per field and draw from it when one begins.

**What the work looks like**

We match bankers in the network to projects like:

- Building a three-statement model and an LBO from a filing set, then writing the rubric a grader uses to mark a model's attempt wrong
- Reviewing model-generated comps and precedent transactions, and explaining where the multiple, the calendarization, or the peer set is off
- Grading model-drafted CIM and pitch material against what an MD would send, and writing the reasoning behind each deduction

Past rates have run $100 to $130 an hour, set per project by scope and depth.

**What we're looking for**

- Professional experience in financial modeling, valuation, or deal execution
- Clear written communication, since most of the work involves explaining your reasoning
- Comfort with ambiguous tasks, and willingness to flag unclear instructions

**Where people in this pool have worked**

The most common backgrounds, though the pool is broader than this:

- Bulge bracket and elite boutique advisory: M&A, leveraged finance, restructuring, ECM/DCM
- Buy-side seats built on banking training: private equity, growth equity, private credit, distressed
- Corporate development and in-house M&A

The firm name matters less than whether you built the models and wrote the materials yourself, recently enough to explain the judgment calls behind them.

**How it works**

Apply here with your resume and confirm where you work from, then take a short AI interview, about 20 minutes total.

We verify your background and add you to the investment banking pool.

When a project needs investment banking expertise, we invite matching people from the pool to that specific listing. It names the rate, the hours, and the client, and any hiring decision is made there.

**What to expect**

This listing does not accept or reject candidates, so you will not get a decision on it. We reach out when a project matches your background. Depending on what our clients are running, that can be within a week or several months.

Finishing the domain expert interview and joining every network you qualify for both improve your odds of being matched. You only need to sign up once to be eligible across all of them.

Investment Banking Expert

We are building a large set of evaluation tasks for the construction and property field. Each task starts from real drawings, documents and site photos, and asks for the same structured deliverable you would produce at work: an egress analysis, a correction list, a plan review decision. You author the task and the correct answer, and a second reviewer in your discipline checks it before it ships.

What you would be building
Tasks drawn from real fire and life safety review work, including: egress and life safety plan check, fire access plan review, fire alarm review from both the design engineer and the fire marshal seat, sprinkler review from the engineer, fire marshal and property insurer seats, hazardous materials control area review, fire occupancy video inspection, firestop penetration photo verification, and fire system inspection, testing and maintenance report review.

Who we are looking for
You currently hold, or recently held, the seat that reviews fire and life safety design: fire marshal or fire prevention plan reviewer, fire protection engineer, or sprinkler and fire alarm designer-reviewer.
Fire marshal appointment, NICET Level III or higher, or PE in fire protection.
3 or more years in the seat, US based, working to IBC, IFC and NFPA standards. We prefer 8 or more years.
You can mark up plan sets and read hydraulic calculations and device layouts.

What makes someone excellent here
The hard part is not citing the standard, it is knowing which occupant load, travel distance or coverage gap gets waved through. Standouts have usually trained junior inspectors or reviewers, written department correction language or an SOP, done acceptance testing as well as plan review, or worked across several NFPA editions and can say what changed and when it applies.

Probably not a fit
Firefighters and fire officers without plan review or inspection responsibility, sprinkler installers without design or review work, alarm technicians who service rather than design systems, and general safety or OSHA specialists.

Paid hourly at $45 to $60 depending on experience, with a path to grow your rate as the project scales.

Fire & Life Safety Reviewer — Vulcan

We are building a large set of evaluation tasks for the construction and property field. Each task starts from real drawings, maps, aerials and site documents, and asks for the same structured deliverable you would produce at work: a plan review comment set, a survey exception schedule, a measured read. You author the task and the correct answer, and a second reviewer in your discipline checks it before it ships.

What you would be building
Tasks drawn from real site, civil and survey work, including: civil grading plan review, erosion and sediment control plan review, stormwater design plan review, roadway plan check, work-zone and traffic control plan review, utility service drawing review, floodplain permit review, civil RFI response review, landscape ordinance and tree protection review, earthwork quantity verification and stockpile volume certification, ALTA Table A survey items, title survey exception review, closing survey setback review, elevation certificate photo checks, topographic map review and map stack reading, historic aerial and Sanborn map dating, and assessor sketch verification.

Who we are looking for
You currently hold, or recently held, the seat that produces or reviews site, civil and survey work: civil engineer, land development engineer, land surveyor, or geospatial or Phase I environmental analyst.
PE in civil engineering, or PLS for the survey and mapping work.
3 or more years in the seat, US based, working to US standards and state DOT or local land development practice. We prefer 8 or more years.
You are comfortable moving between drawing sets, aerials, topographic maps and parcel data.

What makes someone excellent here
The hard part is not reading the sheet, it is knowing which quantity, contour or exception a reviewer takes on faith. Standouts have usually reviewed submittals for an agency or DOT, mentored EITs or survey crews, carried a project through agency resubmittal cycles, or worked across several jurisdictions and can name what each one checks differently.

Probably not a fit
GIS analysts who have not reviewed engineered drawings, drone pilots without survey credentials, environmental scientists whose work is sampling rather than site plan review, and CAD technicians without design or review responsibility.

Paid hourly at $45 to $60 depending on experience, with a path to grow your rate as the project scales.

Civil / Survey Engineer — Vulcan

We are building a large set of evaluation tasks for the construction and property field. Each task starts from real drawings, documents and site photos, and asks for the same structured deliverable you would produce at work: an RFI response, a coordination comment set, a marked-up set. You author the task and the correct answer, and a second reviewer in your discipline checks it before it ships.

What you would be building
Tasks drawn from real design and coordination work, including: architectural RFI response review, architectural set QC check, technical peer review, cross-sheet coordination review, constructability review, redline back-check, design change approval, addendum change scoping, as-built drawing production, pre-bid drawing scope review, drawing-set document control, commissioning design review, and shop drawing review for interiors, millwork and building envelope details.

Who we are looking for
You currently hold, or recently held, the seat that produces and coordinates permit and construction drawing sets: architect of record, project architect, job captain, or design coordination lead.
Licensed architect. In lieu of a license, 8 or more years as a project architect or job captain producing permit sets.
3 or more years in the seat, US based, working in US model code jurisdictions. We prefer 8 or more years.
You can mark up drawing sets and details. Bluebeam Revu or equivalent.

What makes someone excellent here
The hard part is not answering the RFI, it is knowing which cross-sheet conflict a rushed reviewer signs off on. Standouts have usually run set QC or back-check on other people's drawings, mentored job captains, maintained office standards or a detail library, or worked both sides of the RFI process as designer and as reviewer.

Probably not a fit
CAD or BIM drafters without authorship of the set, renderers and visualization specialists, interior decorators, and PMs who coordinate schedules rather than drawings.

Paid hourly at $45 to $60 depending on experience, with a path to grow your rate as the project scales.

Architect — Design Coordination (Vulcan)

We are building a large set of evaluation tasks for the construction and property field. Each task starts from real drawings, documents and site photos, and asks for the same structured deliverable you would produce at work: a correction letter, a review decision, a marked-up set. You author the task and the correct answer, and a second reviewer in your discipline checks it before it ships.

What you would be building
Tasks drawn from real permit review work, including: building plans examiner review, code consultant plan review, third party and private provider plan review, accessibility and ICC A117.1 review, energy code drawing checks, occupancy classification and storefront change of use, zoning site plan review, permit intake completeness screening, as-built versus permit deviation, and specialty occupancy review for healthcare facilities, schools, food service and public pools.

Who we are looking for
You currently hold, or recently held, the seat that reviews permit drawing sets for code compliance: a building department plans examiner, a third party or private provider plan reviewer, or a code consultant.
ICC Building Plans Examiner certification (B1, B2 or B3), or current or recent employment doing AHJ or third party plan review. There is no degree requirement for this role.
3 or more years in the seat, US based, working in US model code jurisdictions (IBC, IEBC, IECC, ICC A117.1). We prefer 8 or more years.
You can mark up and measure a plan set. Bluebeam Revu or equivalent.

What makes someone excellent here
The hard part is not applying the code, it is knowing exactly what a careless reviewer would miss. Standouts have usually trained junior examiners, written or maintained a comment library or correction template, done QA on other reviewers' work, or worked across several code editions and jurisdictions and can say precisely what changes between them.

Probably not a fit
General contractors and superintendents without plan review responsibility, estimators, CAD drafters, and permit expediters who file applications but do not review them.

Paid hourly at $45 to $60 depending on experience, with a path to grow your rate as the project scales.

Code Plan Reviewer (Plans Examiner) — Vulcan

## About the work

We are building expert-authored forecasting and research data used to evaluate and improve AI models on real financial analysis. Each task asks a practitioner to reason from a fixed evidence cutoff — no hindsight — and document the judgment a strong analyst would have made at that moment.

This role covers company event propagation: identifying the companies related to an upcoming event and estimating how the catalyst should transmit to each of them.

## What you will do

- Map suppliers, customers, competitors, distributors, substitutes and geographic exposure from EDGAR filings, footnotes, exhibits, transcripts and industry data
    
- Identify related companies ahead of an event and separate direct exposure from weak thematic correlation
    
- Estimate direction, magnitude, horizon and confidence, and name the revenue, margin, earnings or valuation mechanism behind each
    
- Review model-generated analyses and grade them against your own standard
    

## Who we are looking for

- A senior sector PM, lead equity analyst, or event-driven investor with deep knowledge of a specific industry
    
- Hands-on experience mapping a company's economic network from primary filings, not from screens or third-party tags
    
- Typically 8+ years of direct experience, or unusually strong evidence of equivalent performance
    

Backgrounds we look at first: top multi-manager and event-driven funds, leading sector teams at major asset managers and banks, and elite specialist research firms — for example Citadel, Millennium, Point72, Balyasny, D. E. Shaw, and top sell-side sector franchises. Brand-name experience is a strong first-pass signal but is not a hard requirement — direct experience and performance on the work sample determine who qualifies.

## Point-in-time discipline

Every task is graded on reasoning from the stated cutoff without hindsight, data leakage, confidential information, or material non-public information.

Company Analysis Expert — Event Propagation

## About the work

We are building expert-authored forecasting and research data used to evaluate and improve AI models on real macro analysis. Each task asks a practitioner to reason from a fixed evidence cutoff — no hindsight — and document the judgment a strong analyst would have made at that moment.

This role covers non-G7 central bank communication. We staff separately by country: a broad global-macro profile is not enough, and we are looking for depth on one specific central bank.

## What you will do

- Interpret statements, minutes, speeches and wording changes from a specific non-G7 central bank in their source language
    
- Compare current language against prior guidance and prevailing market expectations
    
- Assign a calibrated dovish-to-hawkish score and spell out the implications for rates, FX and equities
    
- Review model-generated analyses and grade them against your own standard
    

## Who we are looking for

- A former economist or adviser from the relevant central bank, or a senior local rates or FX strategist who covered it through multiple policy cycles
    
- Professional fluency in the source language
    
- Deep knowledge of that bank's reaction function and communication style, and experience reading it in real time
    
- Typically 8+ years of direct experience, or unusually strong evidence of equivalent performance
    

Backgrounds we look at first: the exact central bank, the BIS, the IMF, leading local institutions, top emerging-markets teams at global banks, and established macro funds. Brand-name experience is a strong first-pass signal but is not a hard requirement — direct experience and performance on the work sample determine who qualifies.

## Point-in-time discipline

Every task is graded on reasoning from the stated cutoff without hindsight, data leakage, confidential information, or material non-public information.

Central Bank Expert — Non-G7 Policy Communication

## About the work

We are building expert-authored forecasting and research data used to evaluate and improve AI models on real financial analysis. Each task asks a practitioner to reason from a fixed evidence cutoff — no hindsight — and document the judgment a strong analyst would have made at that moment.

This role covers single-name equity analyst publications: predicting when a named sell-side analyst will publish after a catalyst, and what that update will contain.

## What you will do

- Forecast the timing and content of a named analyst's post-catalyst research note
    
- Reason from filings, earnings, transcripts, estimates, valuation and ratings to an expected published view
    
- Document the internal research workflow that determines how and when an equity research report gets produced
    
- Review model-generated analyses and grade them against your own standard
    

## Who we are looking for

- A former lead sell-side equity analyst or highly senior research associate who personally published post-catalyst research in the exact sector
    
- Deep command of filings, earnings, transcripts, estimates, valuation, ratings, and the internal workflow behind an equity research report
    
- Typically 8+ years of direct experience, or unusually strong evidence of equivalent performance
    

Backgrounds we look at first: Institutional Investor-ranked teams and leading research franchises such as Goldman Sachs, Morgan Stanley, JPMorgan, Bank of America, Citi, UBS, Barclays, Bernstein, Jefferies and Wolfe Research. Former colleagues, direct coverage rivals, and senior buy-side consumers of a named analyst's work are especially valuable. Brand-name experience is a strong first-pass signal but is not a hard requirement — direct experience and performance on the work sample determine who qualifies.

## Point-in-time discipline

Every task is graded on reasoning from the stated cutoff without hindsight, data leakage, confidential information, or material non-public information.

Equity Research Expert — Single-Name Analyst Publications

## About the work

We are building expert-authored forecasting and research data used to evaluate and improve AI models on real political and financial analysis. Each task asks a practitioner to reason from a fixed evidence cutoff — no hindsight — and document the judgment a strong analyst would have made at that moment.

This role covers political report authorship: elections, polling, and political-risk events.

## What you will do

- Author point-in-time political reports and forecasts from a stated evidence cutoff
    
- Make and document directional judgments about polling, win probabilities, and market-implied odds
    
- Explain the causal mechanism behind an event's expected political and market impact
    
- Review model-generated analyses and grade them against your own standard
    

## Who we are looking for

- A senior elections forecaster, campaign strategist, polling or statistical-methodology expert, or political-risk analyst
    
- A record of making live judgments about polling, win probabilities, or odds — not retrospective commentary
    
- Strong polling literacy and causal event analysis
    
- Typically 8+ years of direct experience, or unusually strong evidence of equivalent performance
    

Backgrounds we look at first: top national polling and election-forecasting organizations, major decision desks, senior analytics teams from presidential, Senate or governor races, and political-risk desks at leading banks, funds, or firms such as Eurasia Group and EIU. Brand-name experience is a strong first-pass signal but is not a hard requirement — direct experience and performance on the work sample determine who qualifies.

## Not a fit

Pure communications, journalism, or academic profiles without demonstrated forecasting, polling, or probabilistic judgment.

## Point-in-time discipline

Every task is graded on reasoning from the stated cutoff without hindsight, data leakage, confidential information, or material non-public information.

Political Expert — Elections Forecasting & Political Risk

## About the work

Mercor is assembling a panel of chemistry and chemical safety experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same casework knowledge that identifies a substance describes how it was produced; the same precursor list that supports an investigation is a shopping list read backwards. A model that refuses a routine identification or casework question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who have done casework — identifying substances and precursors, and defending the conclusion in a report or on the stand. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same precursor knowledge that identifies a clandestine route is the route. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Controlled substance casework in a forensic laboratory — identification and quantitation
- Clandestine laboratory response, precursor identification and scene chemistry
- Forensic toxicology — postmortem or human performance casework
- Seized-drug and precursor trend monitoring or intelligence analysis
- Expert testimony and report writing on chemical casework

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Forensic / Toxicological Chemist

## About the work

Mercor is assembling a panel of chemistry and chemical safety experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same dispersion model that sites a plant and sets an evacuation radius also predicts an exposure; the same hazard analysis that finds a failure path is a list of failure paths. A model that refuses a routine exposure-limit question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who have assessed real exposures and real process hazards, and signed off on the controls. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same release model that protects a community estimates a consequence. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Certified industrial hygienist — exposure assessment, sampling strategy, control banding
- Process hazard analysis — HAZOP, LOPA, what-if studies, management of change
- Toxic release and dispersion modelling — consequence analysis, AEGL/ERPG criteria, facility siting
- Ventilation, containment and PPE programme design for chemical hazards
- Regulatory compliance — PSM, Seveso, COMAH, OSHA or national equivalents

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Industrial Hygienist / Process Safety Engineer

## About the work

Mercor is assembling a panel of chemistry and chemical safety experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same method that detects a hazardous compound at trace level defines the threshold someone would try to stay beneath; the same library entry that identifies an unknown tells you what a screen will and won't see. A model that refuses a routine method-validation question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for analysts who have built and validated methods for hazardous compounds — people who know what a detection limit really means and what it implies. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same detection method that finds a compound describes how to evade the screen. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Detection and identification of hazardous compounds — GC-MS, LC-MS/MS, IR, NMR
- Method development and validation for trace-level analysis
- Environmental, occupational or emergency-response sampling and analysis
- Reference standards, spectral libraries and unknown identification
- Laboratory accreditation, QA/QC and proficiency testing

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Analytical Chemist

## About the work

Mercor is assembling a panel of chemistry and chemical safety experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same route design that scales a medicine underpins a controlled compound; the same impurity profile that proves a clean process identifies how something was made. A model that refuses a routine scale-up or kinetics question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for chemists who have actually run routes and scaled them — bench to pilot to plant — and who know which requests are ordinary process chemistry and which are asking for something else. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same route knowledge that shortens a synthesis serves whoever is shortening it. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Synthetic route design and development in pharmaceutical, agrochemical or fine chemicals
- Process chemistry and reaction engineering — optimisation, kinetics, impurity control
- Scale-up from bench to pilot to plant, with the safety review that goes with it
- Precursor sourcing, controlled-substance registration or regulatory chemistry
- Method development and troubleshooting in a GMP or equivalent quality environment

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Synthetic / Process Chemist

## About the work

Mercor is assembling a panel of chemistry and chemical safety experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same toxicology that sets a countermeasure dose describes an exposure someone might intend; the same decontamination data that protects responders tells you what defeats it. A model that refuses a routine countermeasure or PPE question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who have worked chemical threats from the defensive side in an institutional setting — countermeasures, protection, detection, verification. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same agent properties that drive a treatment protocol drive a threat assessment. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Chemical defence research at USAMRICD, DoD, CDC or an allied national institute
- Toxicology of chemical warfare agents and medical countermeasure development
- Decontamination, protection and detection research for chemical threats
- CWC-related work — verification, destruction, designated laboratory analysis
- Consequence assessment and medical management of chemical exposure

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Chemical Defense Researcher

## About the work

Mercor is assembling a panel of energetic materials and propulsion experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same design that breaks rock for a highway cut is a description of directed energy release; the same magazine and transport rules that keep a crew safe are a map of where material sits and how it moves. A model that refuses a routine vibration-compliance question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for licensed practitioners who handle, store and move specialty materials as a daily responsibility, and who sign for them. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same inventory and transport detail that satisfies an auditor is what a diversion would exploit. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Licensed blaster or blasting engineer — mining, quarrying, construction or demolition
- Blast design and vibration control — timing, burden, seismograph monitoring, regulatory limits
- Magazine management, inventory reconciliation and transport of specialty materials
- Storage compatibility, security and regulatory compliance for a commercial licensee
- Training, certification or auditing of blasting crews

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Commercial Blasting Engineer

## About the work

Mercor is assembling a panel of energetic materials and propulsion experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same propulsion analysis that qualifies a motor for flight describes a capability; the same safe-and-arm logic that keeps a system inert is the part someone would want to defeat. A model that refuses a routine performance or qualification question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

This is the engineering half of the domain: we're looking for people who have designed, qualified or tested propulsion and initiation systems, with real hardware and real test data behind them. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same effects model that sizes a safety standoff also sizes an intent. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Munitions or warhead design and effects engineering
- Solid, liquid or hybrid propulsion system design, test and qualification
- Terminal effects modelling — blast, fragmentation, penetration
- Fuzing, safe-and-arm and initiation system engineering
- Test range engineering and instrumentation for propulsion or ordnance

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Propulsion & Effects Engineer

## About the work

Mercor is assembling a panel of energetic materials and propulsion experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same residue analysis that identifies what was used is a description of what to use; the same diversion indicators that catch a bad licensee tell someone how not to be caught. A model that refuses a routine forensic identification question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who have worked scenes and casework under a national certification — identifying materials after the fact, and inspecting the licensees who hold them. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same residue signature that closes a case describes a recipe in reverse. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Federally certified explosives specialist or national equivalent — scene and materials analysis
- Forensic laboratory identification of energetic materials and post-blast residues
- Post-incident analysis and reporting for an agency or laboratory
- Regulatory inspection of licensees and permittees — storage, records and diversion indicators
- Expert testimony or casework on energetic materials

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Federal Certified Materials Specialist / Incident Analyst

## About the work

Mercor is assembling a panel of energetic materials and propulsion experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same formulation work that qualifies a propellant or a display composition sits next to something nobody should be helped with; the same sensitivity data that protects a production line reads as a specification from the other side. A model that refuses a routine thermal-stability question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

This is the chemistry half of the domain: we're looking for people who have formulated, synthesised and tested specialty energetic materials in a licensed or government facility, not modelled them from a distance. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same sensitivity result that sets a safe handling limit is what someone else would want to design around. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Formulation chemist for pyrotechnics, propellants or specialty energetic materials
- Synthesis and characterisation of energetic compounds in a licensed or government facility
- Performance and sensitivity testing — calorimetry, thermal stability, small-scale safety testing
- Scale-up and process safety for energetic material production
- Solid or liquid rocket propellant development — binders, oxidisers, burn-rate modifiers

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Specialty Materials Chemist — Pyrotechnics or Propulsion

## About the work

Mercor is assembling a panel of energetic materials and propulsion experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same assessment that tells a technician how to approach a device tells someone else how it was meant to work; the same compatibility data that keeps a magazine safe describes what stays stable elsewhere. A model that refuses a routine storage or render-safe question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for public safety professionals who have assessed real devices — the render-safe and post-incident side, held under certification and command accountability. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same assessment that makes a device safe explains how it was built. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Certified bomb technician or hazardous device operator — device assessment and render-safe procedures
- Military EOD with civilian public-safety, training or instruction experience
- Bomb squad commander or training officer — team qualification, equipment and protocols
- Post-blast investigation and device reconstruction
- CBRNE technical response — standoff assessment, robotics and disruption tools

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Hazardous Device Specialist / Public Safety Professional

## About the work

Mercor is assembling a panel of nuclear materials and safeguards experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same control list that stops a shipment tells a procurement agent what to avoid declaring; the same pathway analysis that informs policy is a roadmap read from the other direction. A model that refuses a routine export-classification question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who have worked proliferation questions professionally — export control, treaty implementation, or programme analysis — and who can tell a compliance question from a circumvention question. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same control list that blocks a transfer also names what to route around. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Export control licensing, classification or enforcement — dual-use lists, catch-all controls, end-user screening
- Treaty and agreement implementation — NPT, CTBT, safeguards agreements, bilateral cooperation agreements
- Proliferation pathway and programme analysis at a laboratory, agency, ministry or research institute
- Sanctions, procurement network or illicit trade analysis
- Open-source and technical analysis of nuclear programmes

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Nonproliferation Analyst

## About the work

Mercor is assembling a panel of nuclear materials and safeguards experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same signature that attributes a sample is the one someone would want to suppress; the same detection limit that catches material moving can be read as a threshold to stay under. A model that refuses a routine measurement-methodology question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who have characterised real nuclear material — detection, attribution, and the analytical work behind a defensible conclusion. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same signature that identifies a source tells you what to hide. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Nuclear forensics casework or international exercise participation — sample characterisation and attribution
- Radiochemical and mass-spectrometric analysis — isotopics, age dating, impurity and process signatures
- Detection and interdiction technology — portal monitors, handheld identification, search operations
- Material signature libraries, reference materials and attribution methodology
- Pre- or post-detonation forensics research at a national laboratory or agency

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Forensic Scientist — Nuclear Materials

## About the work

Mercor is assembling a panel of nuclear materials and safeguards experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same material-balance calculation that closes an inspector's account can be used to hide what is missing from it; the same protection assessment that hardens a facility maps its weaknesses. A model that refuses a routine material-accounting question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who have been responsible for keeping nuclear material where it belongs — material control and accountability, physical protection, and the assessments that test both. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same vulnerability assessment that closes a gap also describes it. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Material control and accountability (MC&A) programme design, operation or assessment at a national laboratory or licensee
- Physical protection system engineering — sensors, barriers, delay, response force, performance testing
- Vulnerability assessment and adversary path analysis for nuclear facilities
- Insider threat and human reliability programme work
- Nuclear security policy, training or international assistance programmes

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Nuclear Security Engineer

## About the work

Mercor is assembling a panel of nuclear materials and safeguards experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same material-balance calculation that closes an inspector's account can be used to hide what is missing from it; the same fuel-cycle step that supplies a power reactor is the one a proliferation pathway runs through. A model that refuses a routine criticality-safety question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who have worked the fuel cycle itself — enrichment, reprocessing, fabrication, criticality safety — and who know which parameters are routine engineering and which ones only matter to someone going somewhere else with them. What matters most is being able to tell a routine professional question from one that's fishing for something else. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Enrichment, conversion or fuel fabrication engineering — process design, operations, throughput
- Reprocessing and back-end fuel cycle — separations, waste streams, spent fuel management
- Criticality safety engineer — criticality safety evaluations, limits and controls, accident analysis
- Fuel cycle systems analysis and modelling — material flows, cycle performance and economics
- Safety analysis or licensing for a fuel-cycle facility

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Nuclear Engineer, Fuel Cycle

## About the work

Mercor is assembling a panel of nuclear materials and safeguards experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same material-balance calculation that closes an inspector's account can be used to hide what is missing from it; the same fuel-cycle step that supplies a power reactor is the one a proliferation pathway runs through. A model that refuses a routine accountancy question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who have verified nuclear material rather than read about it — inspection regimes, accountancy, and the judgment calls that come with an unresolved discrepancy. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same material-balance figure that closes an account can be used to conceal a diversion. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- IAEA, Euratom or state-system (SSAC/RSAC) inspector — inspection planning, verification, reporting
- Material accountancy — facility declarations, inventory change reports, MUF evaluation and resolution
- Additional Protocol implementation — declarations, complementary access, information analysis
- Safeguards instrumentation and containment/surveillance — NDA measurement, seals, unattended monitoring
- Safeguards-by-design work or safeguards policy analysis at a laboratory, regulator or ministry

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Safeguards Analyst / IAEA or Euratom Inspector

## About the work

Mercor is assembling a panel of radiological safety experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same isotope underpins cancer therapy and industrial radiography; the same shielding calculation protects a worker and conceals a source. A model that refuses a routine dose question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who produce and handle medical isotopes — the legitimate-use half of the dual-use line, held as a real custody responsibility. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same activity and shielding figures that let a patient go home describe a source someone else wants to move quietly. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Nuclear medicine or radiopharmacy physicist — therapy dosimetry, I-131, Lu-177 and Y-90 work, patient release criteria
- Cyclotron or radiochemistry facility RSO — target handling, PET isotope production, hot-cell operations
- Medical physicist with brachytherapy source custody — HDR and LDR sources, inventory and source exchange
- Radiopharmaceutical production and QC under GMP or an equivalent quality system
- Authorised Medical Physicist or Authorised User support on a medical radioactive materials licence

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Nuclear Medicine Physicist / Cyclotron RSO

## About the work

Mercor is assembling a panel of radiological safety experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same isotope underpins cancer therapy and industrial radiography; the same shielding calculation protects a worker and conceals a source. A model that refuses a routine dose question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who have planned for or responded to a radiological event — the consequence, contamination and protective-action side, held as a real responsibility rather than a tabletop interest. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same dispersal model that drives an evacuation decision also describes an effect someone might want. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Radiological emergency planning at a licensed facility, or at state or national level — plans, drills, exercises
- Field response and monitoring — FRMAC, RAP, CBRN or national radiological response teams
- Consequence and dispersal modelling — plume and deposition estimates, dose projection, protective action decisions
- Incident command or response coordination for a radiological event or full-scale exercise
- Medical and public-health radiological response — triage, countermeasures, population monitoring

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Emergency Response Specialist

## About the work

Mercor is assembling a panel of radiological safety experts to red-team frontier AI models. The goal is to test whether a model can correctly judge the misuse potential of a technical request — answering legitimate questions fully while refusing genuinely dangerous ones.

You will:

- Write challenging single-turn prompts in your domain, labelled across three levels: benign, dual-use, and adversarial.
- Evaluate the responses that come back against a defined policy standard, and judge whether each was handled correctly.
- Write the reference answer — what a correct response looks like, and the technical reasoning for why.

This domain lives on the dual-use line more than most. The same isotope underpins cancer therapy and industrial radiography; the same shielding calculation protects a worker and conceals a source. A model that refuses a routine dose question is failing as badly as one that answers a question it shouldn't. Drawing that line reliably is why we need practitioners rather than generalists.

## Background we're looking for

We're looking for people who have assessed how Category 1 and 2 sources actually get stolen, shielded or moved — the security and vulnerability side of a licensed programme, not just its paperwork. What matters most is being able to tell a routine professional question from one that's fishing for something else: the same shielding calculation protects a worker and conceals a source. You will be writing prompts that sit exactly on that line, then judging whether the model held it.

Examples of relevant backgrounds (ideally with red-teaming experience):

- Security or vulnerability assessment of licensed facilities holding Category 1 and 2 sources
- Physical protection design and review — detect, delay, assess and respond; insider threat considerations
- Source recovery, consolidation or security-upgrade programmes for disused and orphan sources
- Regulatory security inspection and compliance — 10 CFR 37 or national equivalents
- Transport security for radioactive material — routing, escort, tracking and chain of custody

## What this role demands beyond technical depth

This is writing-intensive work. Every judgment you make needs a written rationale that a non-specialist can follow. Prior technical writing, published research, or expert witness experience is a strong signal; please include a sample or link. You will also be reading and writing about misuse scenarios in your field for sustained periods. We brief experts on this in advance, and you can pause or step away at any point without penalty.

## Before you apply

Your work here will not involve, and must not draw on, classified or export-controlled information, or anything covered by an NDA or prepublication review obligation. If you hold such obligations you may still be a good fit; tell us in your application and we will scope the work accordingly.

Source Security Specialist

Explore remote opportunities with top companies worldwide. Find roles that match your skills and work preferences.

Explore opportunities