We're hiring experienced inpatient Registered Nurses (RNs) to help train and evaluate AI systems used in clinical and healthcare settings. This role is ideal for nurses who want to apply their frontline experience to improve the accuracy, safety, and reliability of medical AI tools.

You'll work on projects that require deep clinical judgment, attention to detail, and the ability to translate real-world bedside documentation practices into structured feedback for AI systems.

**Key Responsibilities**

- Review and evaluate AI-generated clinical outputs based on nursing documentation
    
- Validate accuracy, completeness, and adherence to current documentation standards
    
- Annotate and structure inpatient nursing assessment data for AI training datasets
    
- Provide expert feedback on nursing assessments and documentation practices
    
- Identify gaps, inconsistencies, or risks in AI-generated responses
    
- Take an ambiguous clinical question, investigate it, and return a thoughtful recommendation
    
- Proactively surface issues, patterns, and opportunities for improvement, and contribute to guideline/taxonomy refinement
    
- Support related clinical product work as needs evolve (annotation, clinical review, model evaluation, user-feedback investigation)
    
- Collaborate with engineers, PMs, and clinical scientists to improve model performance
    

**Basic Qualifications**

- Active RN license (U.S., outside California)
    
- Recent acute-care inpatient bedside RN experience within the last 5 years (current documentation experience is prioritized over total years of nursing experience)
    
- Broad clinical exposure: comfortable reviewing documentation across different inpatient populations, not overly specialized
    
- Comfortable performing and documenting comprehensive nursing assessments
    
- Comfortable following detailed annotation guidelines consistently
    
- Self-starter with high ownership who can work independently and meet deadlines
    
- Excellent written communication skills and responsiveness to feedback
    
- Able to reliably and consistently dedicate a minimum of 10 hours per week
    

**Preferred Qualifications**

- Experience at a large academic or teaching health system
    
- Previous data annotation, chart abstraction, or healthcare-AI experience with positive performance reviews
    
- Epic EHR experience (most common platform, closely aligns with many workflows)
    
- Comfortable learning new annotation tools and web-based platforms
    
- Curious and adaptable: interested in how the product/model works and comfortable with processes that change quickly

RN Annotators

**About the work**

We're working on a project partnering with a leading AI lab to improve how AI models work with

technical and scientific data related to frontier STEM fields! We need STEM experts like you to help us complete this work, utilizing your expertise!

**Who we're looking for**

Graduate-level training or extensive industry experience in one of the following:

- Engineering
    
- Physics
    
- Mathematics
    
- Chemistry
    
- Software engineering / data science
    

Experience working with data and data visualization is preferred.

Strong written English, and the judgment to verify a result against a source rather than eyeball it.

**Details**

- Minimum 5 hours per week.
    
- Applications reviewed on a rolling basis.

STEM Expert

## 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

# **Domain Expert – FP&A (Financial Forecaster)**

**To qualify you must use Financial Forecaster regularly, weekly or more, as a working part of your job, with 5+ years of professional FP&A experience as an FP&A analyst/manager, financial reporting or technical accounting analyst, lender-reporting or treasury analyst, or controller.**

### Financial Forecaster knowledge we require

**Working the model**

- Multidimensional models: navigating models, dimensions and account hierarchies, and reading a cube slice at an exact coordinate rather than at whatever grain the UI lands on
    
- Scenarios: creating one, setting its parent, knowing what it inherits versus holds in its own cells, and locking it — plus identifying the forecast of record by its lock or acceptance date
    
- Accounts and assumptions: telling an input account from a calculated one, and knowing when to fix a figure by editing the account formula versus overriding an assumption cell, because the two are not interchangeable and the grader can tell
    
- CSV imports: loading actuals and driver data, and the unit and scale discipline an import demands
    
- Note surfaces: scenario notes, driver notes, policy notes, planning notes, assumption change logs and variance commentary — and which one a given statement belongs in
    
- Knowing what Financial Forecaster cannot do, so you can recognize a task that asks for a figure or a dimension member the model will never produce
    

**Reading models & packs**

- Reading a scenario fluently: what is an input versus a calculated account, which cells are overridden, what a scenario inherits from its parent versus holds itself
    
- Reading a reporting pack against the model behind it: tracing a printed figure back to the account, scenario and coordinate it came from, and telling a real tie-out from a coincidence
    
- Recognizing the vintage that governs: which scenario is the forecast of record, when it was locked or accepted, and why a later forecast does not silently supersede an approved one
    
- Decoding a coordinate (Time, Entity, Department, Product/Product Line, Region, and facility dimensions like Debtor or Aging Bucket) and knowing what a figure means at that grain versus aggregated
    

**Finding & tying out the number**

- Navigating account hierarchies and scenario trees to locate the figure that actually drives a covenant, a KPI, or a disclosure — targeted retrieval instead of scrolling the model
    
- Unit and scale traps: thousands-versus-dollars mismatches, a cost base that cannot be reconciled to its own revenue, per-period versus LTM versus fiscal-year-to-date, and sniffing out a figure that is off by 10× or 1,000×
    
- Reconciling two in-world sources that disagree, and judging which one governs
    

**Entity structure & governing documents**

- The reporting landscape: legal entity versus department versus segment versus consolidated, and which level a given obligation is measured at
    
- Credit agreement mechanics: the provision fixing the accounting basis (frozen GAAP), Test Periods and test dates, fixed charge coverage and total net leverage, Consolidated EBITDA add-backs and their caps, Consolidated Funded Indebtedness, borrowing base and Eligible Accounts, redetermination, and who must sign a compliance certificate
    
- Case relationships in the financial sense: parent and child scenarios, consolidated versus entity results, and how an amendment, consent or waiver changes the measure going forward
    

**Measure types & classification**

- GAAP versus non-GAAP versus covenant measures, and why the same label can mean three different numbers in one pack
    
- Non-GAAP discipline: exclusions requiring a written determination that an amount is not normal and recurring, reconciliation to the nearest GAAP measure, equal prominence, and restating prior periods when a metric definition changes
    
- Forecast versus projection under the AICPA PFI guidance, range presentations whose endpoints must be comparably likely, and what a reasonably objective basis requires
    
- The standards that decide the answer: going concern and the look-forward window (ASC 205-40), operating segments and the CODM (ASC 280), materiality screening (SAB 99 / Topic 1M), KPI disclosure (Release 33-10751), known trends in MD&A (Reg S-K Item 303), valuation allowances as a discrete item versus through the annual effective rate
    

**Documents, records & workflow**

- Document types: compliance certificate, borrowing base certificate, lender package, board or committee pre-read, earnings release and reconciliation, disclosure committee memo, workpaper, and what each is for
    
- What a finished deliverable requires versus what merely describes it: a correction stated in a memo is not the same as the correction applied in the file or written to the model
    
- App workflow surfaces: scenario creation and locking, saved views, and exports for large result sets
    

## What you'll do

- **Review and QA AI-agent rollouts:** work through completed agent runs in Financial Forecaster and judge whether the agent's answer is correct against the actual model and documents — right scenario, right account, right coordinate, right basis, right conclusion, right deliverable
    
- **Identify failure modes:** pinpoint where and why an agent succeeds too easily or fails wrongly, and flag tasks that are too easy or solvable without actually using Financial Forecaster
    
- **Improve realism and difficulty:** refine the task prompt and the seeded world so tasks reflect authentic FP&A and lender-reporting work and are genuinely hard for a capable agent
    
- **Update grading guidance for accuracy:** sharpen how answers are graded so correct answers pass and wrong ones fail, including gating failures (wrong scenario or vintage used, wrong accounting basis, an add-back taken that the agreement excludes, a figure stated but never written to the model, or a conclusion transcribed from a document that already gave it away)
    
- **Catch plausible-but-wrong answers** that a non-expert reviewer would let slide — a number that looks right at the wrong grain, or a conclusion that is correct in isolation and inconsistent with the rest of the pack
    

## Requirements

- Frequent hands-on Financial Forecaster use for real forecasting, covenant or reporting work in your current or recent role, weekly or more
    
- 5+ years in FP&A, financial reporting, technical accounting, or lender reporting: FP&A analyst/manager, financial reporting analyst, technical accounting analyst, treasury/lender-reporting analyst, or controller
    
- Can read a Financial Forecaster model on sight: tell inputs from calculations, spot an override, identify the governing scenario and its vintage, and judge whether a figure ties at the grain it is stated at
    
- Reflexive with Financial Forecaster's scenario and import mechanics, and with the scale and basis traps that produce confidently wrong numbers
    
- Sound judgment on what a correct, complete answer looks like, and the confidence to mark a plausible answer wrong
    
- **Bonus:** credit agreement compliance and borrowing base reporting; going concern or impairment analysis; segment reporting; SEC reporting or disclosure committee experience; hands-on experience across multiple models and entities, not just one company's single model
    
- **Note:** a CPA, CFA, or MBA is welcome but not required; experienced FP&A analysts and technical accounting specialists are the primary pool
    

* * *

FP&A Expert

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)

**Role overview**

Mercor is hiring power systems engineers with recent hands-on industry experience to review and annotate complex electrical engineering tasks. You will assess realistic design and study scenarios and produce clear written technical findings.

**What you will do**

- Review tasks across substation design, transmission design, and generation interconnection: circuit and control design, switchgear and protection, single-line diagrams, and power distribution
    
- Check engineering assumptions, calculations, code and standards use, and equipment data
    
- Flag ambiguity, missing constraints, and unrealistic deliverables
    
- Produce clear written findings
    

**Requirements**

- Recent hands-on experience as a power systems professional in one or more of: substation design, transmission design, or generation interconnection. This is mandatory.
    
- Minimum 1 year of relevant industry experience
    
- Circuit design depth: protection and controls, switchgear, power distribution, and the associated calculations
    
- Experience at an EPC contractor, utility, transmission owner, power generation developer, ISO/RTO, or owner's engineering group
    
- Your resume should clearly show relevant power systems experience
    

**Details**

- Approximately 10 to 20 hours per week initially
    
- Immediate start
    

We consider all qualified applicants without regard to legally protected characteristics and provide reasonable accommodations upon request.

Power Systems Engineer (Substation, Transmission, Interconnection)

## About the Role

Mercor is hiring **residency-trained physicians** across specialties for non-clinical work developing and evaluating clinical AI systems. You will apply your clinical judgment to grading-criteria development, dialogue evaluation, and structured annotation work that determines how these systems are measured.

This is a **non-clinical role** — no direct patient care, and no responsibility for live diagnosis.

**This is a shared expert pool.** After onboarding you may be matched to any of several concurrent clinical workstreams based on your specialty, availability, and interest. You are not committing to a single project, and you may move between streams as priorities shift.

## What you may work on

Work varies by workstream and may include:

- **Grading criteria development** — taking a clinical question and breaking the ideal answer into discrete, checkable criteria, so a model response can be graded consistently rather than impressionistically.
    
- **Clinical dialogue evaluation** — reviewing multi-turn clinical conversations and judging them for accuracy, safety, completeness, appropriate hedging, and whether escalation advice was correct.
    
- **Clinical reasoning annotation** — recording how you would work through a case, including the differential you considered and rejected, not only the conclusion.
    
- **Output review** — flagging hallucinated findings, dangerous omissions, unsupported certainty, and advice that is technically correct but clinically unsafe.
    
- **Guideline authoring** — defining edge cases and standards of care for your specialty so annotation stays consistent across a large group of clinicians.
    
- **Difficult-case writing** — constructing clinical questions that probe the limits of current model reasoning.
    

Task length varies by stream, from roughly 45 minutes for a dialogue evaluation up to an hour or more for grading-criteria authoring. You will get a specific throughput target for whichever stream you are matched to.

## Required qualifications

- **MD or DO** with a **completed residency** in any specialty
    
- **Active, unrestricted medical license** in your country of practice
    
- **3+ years post-residency** clinical experience, practising or previously practising
    
- Comfort writing structured clinical rationale that a non-specialist reviewer can follow
    
- Written and spoken English fluency
    
- **Minimum 20 hours per week**, with the ability to concentrate hours when a stream is time-boxed
    

## Preferred qualifications

- Board certification in your specialty
    
- **U.S. licensure**, and familiarity with U.S. standards of care and clinical guidelines
    
- Primary care, internal medicine, emergency medicine, or hospitalist background, where breadth of presentation matters most
    
- Prior clinical annotation, AI evaluation, medical education, or question-writing experience
    
- Grading-criteria design, resident assessment, or clinical guideline development experience
    
- Published research or sustained technical writing (please link a sample)
    

## Why this work

Most clinical AI failures are not exotic — they are ordinary questions answered with misplaced confidence. Catching that requires someone who has actually carried clinical responsibility. The standard these systems get held to is written by physicians, and here you would be writing it.

General Clinician (MD/DO) - HLS Expert Pool

**Role Overview**

Mercor is collaborating with a leading AI lab to engage experienced accounting professionals across all areas of practice — including audit, tax, financial reporting, bookkeeping, controllership, forensic, and accounting systems. Contributors help build AI systems that reason about real accounting work by translating everyday accounting workflows, judgments, and decision-making into structured, high-quality training data.

**Key Responsibilities**

- Design realistic accounting scenarios and tasks drawn from your day-to-day work (e.g., financial statement preparation, reconciliations, journal entries, audit procedures, tax filings, month-end close, internal controls)
    
- Review and compare AI-generated accounting outputs for accuracy, standards compliance (GAAP/IFRS), and sound professional judgment
    
- Create structured examples that reflect how accountants actually reason through problems
    
- Provide clear written feedback that improves how AI performs accounting tasks
    
- Collaborate asynchronously with the research team
    

**Ideal Qualifications**

- 3+ years of professional experience in accounting, audit, tax, bookkeeping, or finance operations (public accounting, corporate/in-house, advisory, or a firm)
    
- CPA, CA, ACCA, CMA, EA, or an equivalent professional credential required
    
- Bachelor's degree in Accounting, Finance, or a related field
    
- Comfortable with common accounting tools (e.g., QuickBooks, NetSuite, SAP, Oracle, Excel)
    
- Strong written communication and attention to detail
    

**More About the Opportunity**

- Open to all accounting specialties — contribute where your expertise is strongest
    
- Work spans task design, evaluation, and structured feedback on AI accounting outputs
    
- Strong contributors advance into reviewer, lead, and domain-expert roles
    

**Application Process**

- Submit a resume or a short summary of your accounting experience
    
- Complete a short form on your practice area, specialties, and certifications
    
- Selected applicants may complete a brief sample task
    
- Follow-up typically provided within a few days

Accounting Expert (CL Funnel)

**About This Survey**

Mercor is opening a series of compliance and financial-crime engagements with leading AI labs over the coming weeks and months — spanning AML and BSA, KYC and client due diligence, sanctions screening, transaction monitoring and financial crime investigation, securities and wealth compliance, conduct and surveillance, regulatory reporting, and compliance testing and assurance. Before we finalize the scope of each, we want to build them around how compliance professionals _actually_ work, not how the field is described in textbooks or public sources.

This is a written intake designed to map out the real workflows, tools, decision points, and specialty cuts inside the profession — from the inside.

**What You'll Do**

- Complete a ~15 minute written intake about your compliance practice
    
- Describe your day-to-day workflows, how your time is split, and the workflows core to your role
    
- Explain how experienced practitioners actually divide the field from the inside — the cuts insiders use, not the textbook ones
    
- Tell us where AI already gets your work wrong, with concrete examples from your own casework
    
- Share the "invisible" parts of your work that outsiders (including project designers) tend to miss
    
- No right or wrong answers — detail and specificity matter far more than polish
    

**Ideal Respondents**

- Practicing or recently practicing compliance and financial-crime professionals — first line, second line, internal audit, or external advisory
    
- Any seat: KYC/CDD and client onboarding, transaction monitoring and AML investigations, sanctions and screening, fraud and disputes, trade and communications surveillance, conduct and consumer-protection compliance, regulatory reporting and change, compliance testing and assurance
    
- CAMS, CGSS, CFCS, CFE, CRCM, an ICA diploma, or FINRA registrations preferred; deep domain experience without a credential is equally welcome
    
- Willing to write detailed, specific responses about the mechanics of your work
    

**Why This Matters**

- Your intake shapes the scope of the compliance engagements we launch
    
- Thoughtful responses translate directly into **more offer matches** and priority routing as we open new engagements over the coming weeks and months
    
- Typical compliance rates on Mercor range from **$90–$110/hr** and scale with seniority
    

**About Mercor**

Mercor is a talent marketplace that connects experts with leading AI labs and research organizations. Backed by Benchmark, General Catalyst, Adam D'Angelo, Larry Summers, and Jack Dorsey.

Financial Compliance Expert (CL Funnel)

**Role Overview**

Mercor is collaborating with a leading AI lab to engage experienced insurance professionals across all areas of practice — including underwriting, claims, actuarial, risk management, brokerage, compliance, reinsurance, and policy/product. Contributors help build AI systems that reason about real insurance work by translating everyday insurance workflows, judgments, and decision-making into structured, high-quality training data.

**Key Responsibilities**

- Design realistic insurance scenarios and tasks drawn from your day-to-day work (e.g., risk assessment and underwriting decisions, claims adjudication, policy interpretation, coverage analysis, actuarial pricing/reserving, regulatory compliance reviews)
    
- Review and compare AI-generated insurance outputs for accuracy, sound professional judgment, and regulatory/standards compliance
    
- Create structured examples that reflect how insurance professionals actually reason through problems
    
- Provide clear written feedback that improves how AI performs insurance tasks
    
- Collaborate asynchronously with the research team
    

**Ideal Qualifications**

- 3+ years of professional experience in insurance (underwriting, claims, actuarial, risk management, brokerage, compliance, or a related area; carrier, broker/agency, reinsurer, or advisory)
    
- A professional credential such as CPCU, ARM, AINS, CLU, AIC, ACAS/FCAS or ASA/FSA (actuarial), or a state adjuster/broker license is preferred
    
- Bachelor's degree in a related field
    
- Comfortable with common industry tools and systems (e.g., policy administration, claims, and rating platforms, Excel)
    
- Strong written communication and attention to detail
    

**More About the Opportunity**

- Open to all insurance specialties — contribute where your expertise is strongest
    
- Work spans task design, evaluation, and structured feedback on AI insurance outputs
    
- Strong contributors advance into reviewer, lead, and domain-expert roles
    

**Application Process**

- Submit a resume or a short summary of your insurance experience
    
- Complete a short form on your practice area, specialties, and certifications
    
- Selected applicants may complete a brief sample task
    
- Follow-up typically provided within a few days

Insurance Expert (CL Funnel)

## Non-Investment Finance Expert — AI Training & Evaluation

We're building frontier AI for complex, real-world finance work, and we need senior finance practitioners to produce and evaluate expert-grade work products that set the bar for what "correct, professional output" looks like. You'll work through realistic, document-heavy scenarios — messy source files, distractors, ambiguous guidance — and either produce the deliverable yourself or judge AI-generated attempts against a professional standard. This is hands-on finance craft, not survey-taking.

### What you'll do

- Build and review deliverables such as: PnL and mark-to-market reconciliations (trader vs. accounting, month-end marks, commissions, adjustments); vendor / AP payment reconciliations under cash constraints with controls (PO, P-card, aging); integrated 3-statement and pro-forma models including purchase-accounting impacts (ASC 805 deferred-revenue haircut, PPA amortization, synergy phasing); covenant and credit analysis; and partnership / professional-services allocations (profit-pool waterfalls, capital accounts).
    
- Reason from primary sources — term sheets, purchase/partnership agreements, trial balances, memos, emails — and separate what's material from deliberate distractors.
    
- Write clear, CFO/partner-ready narratives that explain assumptions, variances, and recommendations.
    

### Who we're looking for

- 3+ years in a hands-on non-investment finance role: product control / financial control, corporate FP&A, technical accounting, treasury / financial operations / AP–P2P, commercial-banking credit, or professional-services / partnership finance.
    
- CPA / CA / ACA / ACCA / CIMA (or equivalent) strongly preferred.
    
- Strong US GAAP / IFRS (bonus: ASC 805 & 606) and advanced Excel modeling (linked 3-statement models, reconciliations, pro formas).
    
- Comfort exercising professional judgment on incomplete or ambiguous information.
    
- Note: a CFA / investing / portfolio-management background is not required — this is operational and technical finance, not markets.

Non-Investment Finance Expert (CL Funnel)

**About the Role**

We're looking for professionals with deep, hands-on experience doing the analytical and decision-making work at the core of finance — the modeling, diligence, valuation, and judgment calls that sit behind real capital, risk, and reporting decisions. You'll design realistic, complex scenarios from your own practice area, build the supporting materials (models, memos, filings, data), then author and evaluate finance tasks used to train AI agents. This is authoring and judgment work, not throughput.

**Priority Areas — we are hiring hardest here**

\- Investment Banking & M&A: valuation (DCF, comps, precedent transactions, LBO), deal execution, fairness analysis, merger models, and diligence

\- Private Equity & Private Credit: underwriting, LBO modeling, capital structure, covenant analysis, and portfolio-company work

\- Corporate Finance & FP&A: three-statement modeling, forecasting and budgeting, variance analysis, capital allocation, and board/management reporting

**Also In Scope**

\- Equity and credit research, buy-side and sell-side coverage

\- Asset and wealth management, portfolio construction, and manager selection

\- Quantitative finance, risk management (market, credit, liquidity), and derivatives pricing

\- Treasury, capital markets, and corporate development

\- Financial modeling and technical accounting where it drives valuation or reporting judgment (US GAAP / IFRS)

\- Real estate, infrastructure, and project finance

\- Regulatory and compliance work in finance (Basel, SEC/regulatory reporting, capital adequacy)

**Requirements**

\- 3+ years of substantive, recent, hands-on finance experience at a recognized institution (investment bank, asset manager, PE/credit fund, Big Four, a large corporate finance function, or a financial regulator), and currently working in the field

\- Direct experience with the tools this work runs on — Excel/financial modeling, and where relevant Bloomberg/Capital IQ/FactSet, data rooms, ERP/consolidation systems, or quant/risk platforms

\- A 10+ hrs/week commitment

**What Makes a Strong Candidate**

We are looking for people who can construct a complicated problem, not people who process volume. The strongest signals are evidence of owning analysis and decisions: building the model that drove a live deal or investment decision, leading diligence, defending a valuation or recommendation to an investment committee or client, owning a forecast and reporting cycle, structuring a financing, or writing the memo others relied on. Within any competency we prefer the senior end (VP, Director, Principal, Managing Director, Portfolio Manager, Controller, CFO), and experience across two or more disciplines — or on both the buy side and sell side — is a strong plus. An advanced degree (MBA, MS, PhD) and/or a recognized credential (CFA, CPA, FRM, actuarial designation) is preferred, not required.

**Settings**

Not bulge-bracket only. Also boutique and middle-market advisory, PE/VC/credit funds, hedge funds and asset managers, corporate finance and treasury functions, family offices, insurers, banks and lenders, fintech, and financial regulators.

Investment Finance Expert (CL Funnel)

**About the Role**

We're looking for professionals with deep, hands-on experience in the administrative and operational work that keeps healthcare organizations running — the "back office" of a medical practice, hospital, or health plan. You'll design realistic, complex scenarios from your own practice area, build the supporting records, then author and evaluate healthcare-operations tasks used to train AI agents. This is authoring and judgment work, not throughput.

**Priority Areas** — we are hiring hardest here

- **Medical Coding:** inpatient, outpatient and pro-fee coding, risk adjustment / HCC coding, coding audit
    
- **Prior Authorization & Utilization:** initiating, submitting and tracking authorizations through payer portals; utilization review and management
    
- **Revenue Cycle Operations:** denials and appeals, A/R follow-up, payment and revenue integrity, underpayment recovery, coordination of benefits and secondary billing
    

**Also In Scope**

- Claims & billing, and full-cycle medical billing
    
- Regulatory & compliance, HIPAA privacy, healthcare internal audit
    
- Payer-side operations: claims adjudication, appeals & grievances, benefit configuration, provider network, utilization management
    
- Coding audit, Clinical Documentation Integrity (CDI), DRG validation, Health Information Management
    
- Revenue integrity, charge master (CDM), managed care and reimbursement analysis
    
- Practice, clinic and hospital administration; Director/VP of Revenue Cycle
    

**Requirements**

- 2+ years of hands-on, recent experience in a healthcare administrative / back-office role, and currently working in one
    
- Direct experience with the systems this work runs on — payer portals (Availity, Optum/Change Healthcare, Waystar, Office Ally, or individual payer hubs), EHR / practice-management platforms, encoders and computer-assisted coding tools, and/or clearinghouses
    
- A 10+ hrs/week commitment
    

**What Makes a Strong Candidate**

We are looking for people who can construct a complicated problem, not people who process volume. The strongest signals are evidence of owning exceptions and disputes: owning an appeal or payer dispute end to end, running coding or DRG audits, defending a code assignment through appeal, leading an EHR conversion or payer implementation, writing SOPs or payer-specific workflow documentation, supervising staff, or sitting on a denials or audit committee. Within any competency we prefer the senior end, and experience across two or more care settings — or on both the provider and payer side — is a strong plus.

**Settings**

Not hospitals only. Also physician groups and ambulatory practices, ambulatory surgery centers, skilled nursing and long-term care, home health and hospice, behavioral health, dialysis, infusion and specialty pharmacy, retail pharmacy and PBM prior authorization, DME suppliers, FQHCs, IDD and waiver services, health plans and TPAs, and RCM outsourcers and clearinghouses.

**Location**

The United States is our primary market. We are also open to candidates in Canada, the United Kingdom, Ireland, continental Europe, Australia and New Zealand. If you are outside the US, you are expected to work in your own country's insurance, funding or billing system — US payer knowledge is a plus, never a requirement.

**Not In Scope**

Clinical care roles (RN, LPN, NP, PA, physician, pharmacist), EMTs and paramedics, medical technicians of any kind (laboratory, radiology, surgical, pharmacy, sterile processing, respiratory, phlebotomy), reception and front-desk check-in, patient registration, appointment scheduling and appointment management at any level, call-centre script-following, medical scribes, and transcription-only roles.

Non-Physician Admin (CL Funnel)

**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)

Mercor is hiring **Castilian Spanish Voice Actors** to work directly with a leading AI lab, improving their model's ability to handle and produce audio data.

**Professional experience in Voice Acting is greatly preferred but not required.** We welcome people with speaking and audio experience through Music, Television, Podcasts, Twitch Streaming, Comedy, YouTube, Radio, or other related fields.

You must be a **native speaker of Castilian (European) Spanish**, born and raised in **Spain**. This role is distinct from our Mexican Spanish listing — please apply to the one matching your native variety.

You also need to be a proficient English speaker to ensure proper communication during the project.

**Project logistics**

- **Participants are paid $250 USD per finished hour**, based on the length of submitted audio.
    
- You'll need to commit between 2-10 hours per week.
    
- Rolling start — onboarding begins once your contract is accepted and ID verification clears.
    
- You will need to be available for a kickoff call.
    

**Requirements**

- Ability to speak in a "customer service" voice — neutral to upbeat tone, clear enunciation, consistent pacing.
    
- Close attention to punctuation, commas and dashes. It is very important that you record transcripts exactly as written, or edit them to reflect natural pacing.
    
- A **professional audio recording setup**: a home studio with a professional-grade microphone, soundproofing, an audio interface, a pop filter, and editing software for post-production.
    
- Submissions must be WAV, 44.1 kHz, mono, with under one second of silence at the start and end, and no audible breathing.

Spanish (Spain) Professional Voice Actor

Mercor is hiring **French-Canadian Voice Actors** to work directly with a leading AI lab, improving their model's ability to handle and produce audio data.

**Professional experience in Voice Acting is greatly preferred but not required.** We welcome people with speaking and audio experience through Music, Television, Podcasts, Twitch Streaming, Comedy, YouTube, Radio, or other related fields.

You must be a **native speaker of Canadian French**, born and raised in **Canada**.

You also need to be a proficient English speaker to ensure proper communication during the project.

**Project logistics**

- **Participants are paid $300 USD per finished hour**, based on the length of submitted audio.
    
- You'll need to commit between 2-10 hours per week.
    
- Rolling start — onboarding begins once your contract is accepted and ID verification clears.
    
- You will need to be available for a kickoff call.
    

**Requirements**

- Ability to speak in a "customer service" voice — neutral to upbeat tone, clear enunciation, consistent pacing.
    
- Close attention to punctuation, commas and dashes. It is very important that you record transcripts exactly as written, or edit them to reflect natural pacing.
    
- A **professional audio recording setup**: a home studio with a professional-grade microphone, soundproofing, an audio interface, a pop filter, and editing software for post-production.
    
- Submissions must be WAV, 44.1 kHz, mono, with under one second of silence at the start and end, and no audible breathing.

French-Canadian Professional Voice Actor

**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.

Finance Expert

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

**Mercor is seeking an experienced physician with expertise in Ulcerative Colitis (UC)** to support a specialised healthcare market research project. This role focuses on translating a structured research brief into a high-quality physician survey designed to generate meaningful insights into the treatment and management of Ulcerative Colitis. The ideal expert combines strong clinical knowledge of UC with prior experience **developing surveys, questionnaires, or discussion guides for healthcare or pharmaceutical market research**.

## Key Responsibilities

- Review a provided research brief outlining target cohorts, research objectives, screening criteria, and key areas of investigation.
    
- Develop a complete **physician-facing Ulcerative Colitis survey** from the research brief.
    
- Translate research objectives into clear, clinically accurate, unbiased, and appropriately structured survey questions.
    
- Incorporate provided cohort definitions and screening criteria into the overall survey flow.
    
- Design relevant questions covering areas such as patient segmentation, disease severity, treatment pathways, therapy selection, treatment switching, clinical decision-making, and unmet needs.
    
- Structure appropriate response options, scales, follow-up questions, skip logic, and other survey elements where required.
    
- Follow provided survey-writing guidelines and existing templates closely to ensure the final deliverable matches the required format and style.
    
- Review the completed survey for clinical accuracy, logical flow, clarity, neutrality, and completeness before submission.
    

## You’re a Strong Fit If You Have

- An **MD/DO or equivalent medical degree**, ideally with a background in gastroenterology or another field involving direct management of Ulcerative Colitis.
    
- Strong familiarity with the **Ulcerative Colitis treatment landscape**, including disease assessment, treatment sequencing, biologics, targeted therapies, remission, and real-world physician decision-making.
    
- Prior experience **writing healthcare or pharmaceutical market research surveys**, questionnaires, discussion guides, or similar research instruments.
    
- Experience working with pharmaceutical companies, healthcare market research firms, medical affairs teams, life sciences consulting firms, or healthcare insights organisations is strongly preferred.
    
- Ability to translate broad research objectives into precise, answerable survey questions and appropriate response options.
    
- Strong understanding of survey design principles, including neutral wording, question sequencing, response scales, screening logic, and avoidance of bias.
    
- Ability to work from established examples and adhere closely to detailed survey-writing guidelines and output requirements.
    
- Excellent written communication, clinical judgment, and attention to detail.
    

## Project Inputs

The expert will receive:

- A detailed **research brief** containing key cohort information, proposed filter questions, and research areas.
    
- A predefined **cohort and screener section** outlining participant inclusion criteria.
    
- Example completed survey templates demonstrating the expected structure, formatting, level of detail, and writing style.
    

## Expected Deliverable

- A **fully drafted Ulcerative Colitis physician survey document**, following the structure and quality standards demonstrated in the provided templates.
    
- The survey should be ready for review and subsequent implementation with minimal additional drafting required.
    

## Why Join

- Contribute directly to high-impact healthcare and life sciences research.
    
- Apply both clinical expertise and research experience to shape professionally deployed research materials.
    
- Work on a focused, clearly scoped project with detailed briefs, templates, and expected outputs provided upfront.
    
- Collaborate remotely and flexibly on specialised projects aligned with your professional expertise.

Ulcerative Colitis Physician – Medical Market Research Survey 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.

Management Consulting 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 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

## 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 done the measurement and the dose planning themselves, in a facility where getting it wrong mattered — survey, assay, contamination control, work planning. 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):

- Operational health physicist at a DOE site, national laboratory or reactor facility — survey, assay, contamination control
- Internal dosimetry or bioassay specialist — intake assessment, dose reconstruction, committed dose
- Decommissioning and D&D health physicist — characterisation surveys, MARSSIM, release criteria
- Accelerator or reactor health physics — activation products, neutron fields, shielding design and verification
- Dose planning for high-hazard work — RWP development, ALARA reviews, hot-cell and glovebox operations

## 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.

Health Physicist

## 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 held the licence, not just worked under it — the person of record for acquisition, inventory, transfer and disposal. 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):

- Named RSO on a broad-scope licence — programme management, source inventory, regulatory compliance
- RSO or deputy RSO at a university, hospital or research institute running a mixed radionuclide and sealed-source programme
- Industrial or irradiator RSO — Category 1 and 2 sealed sources, radiography, gauging, sterilisation
- Licensing and compliance specialist — NRC, Agreement State or national regulator submissions, amendments and inspections
- Radiation protection manager accountable for ALARA, audits, training and incident reporting across multiple authorised users

## 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.

Radiation Safety Officer

**Help make the next generation of AI-narrated audiobooks sound genuinely human.**

## 1. Overview

We're building a team of native Italian speakers who love audiobooks to help evaluate AI-narrated (text-to-speech) audiobooks in Italian (Italy). You'll listen the way a real audiobook fan does — but with a careful, analytical ear — and flag the moments where the narration slips, so the underlying models can be made better.

This is remote, part-time work at roughly 20 hours per week (about 4 hours per day). You can be based anywhere, though we prefer people living in a region where Italian is natively spoken.

## 2. What You'll Do

- Listen to AI-narrated audiobooks in Italian and assess how natural, accurate, and enjoyable the narration is.
- Spot and log errors in the synthesized speech — skipped, added, or mispronounced words; numbers read incorrectly; abbreviations expanded wrongly; unnatural phrasing or intonation; audio glitches and artifacts.
- Categorize each issue and mark the exact span where it occurs, using our review tooling.
- Report on overall listener satisfaction and where the narration feels natural versus where it breaks the experience.
- Work consistently and carefully across many hours of audio without losing attention to detail.

## 3. Who We're Looking For

- Native or near-native fluency in Italian (Italy), and an active, genuine audiobook listener — someone who listens regularly and has real opinions about what makes a good listen.
- Enough working English to report clearly and communicate with the team (you do not need to be bilingual).
- Some experience with annotation, labeling, transcription, proofreading, linguistics, or other detail-oriented review work.
- A sharp, patient ear and the discipline to stay accurate over long listening sessions.
- Reliable availability of about 20 hours per week on a part-time basis.

## Nice to Have

- Prior experience on AI training-data, model-evaluation, or audio/QA annotation projects.
- A background in language, editing, voice, or audio production.

Audiobook QA Expert — Italian

**Help make the next generation of AI-narrated audiobooks sound genuinely human.**

## 1. Overview

We're building a team of native Dutch speakers who love audiobooks to help evaluate AI-narrated (text-to-speech) audiobooks in Dutch (Netherlands). You'll listen the way a real audiobook fan does — but with a careful, analytical ear — and flag the moments where the narration slips, so the underlying models can be made better.

This is remote, part-time work at roughly 20 hours per week (about 4 hours per day). You can be based anywhere, though we prefer people living in a region where Dutch is natively spoken.

## 2. What You'll Do

- Listen to AI-narrated audiobooks in Dutch and assess how natural, accurate, and enjoyable the narration is.
- Spot and log errors in the synthesized speech — skipped, added, or mispronounced words; numbers read incorrectly; abbreviations expanded wrongly; unnatural phrasing or intonation; audio glitches and artifacts.
- Categorize each issue and mark the exact span where it occurs, using our review tooling.
- Report on overall listener satisfaction and where the narration feels natural versus where it breaks the experience.
- Work consistently and carefully across many hours of audio without losing attention to detail.

## 3. Who We're Looking For

- Native or near-native fluency in Dutch (Netherlands), and an active, genuine audiobook listener — someone who listens regularly and has real opinions about what makes a good listen.
- Enough working English to report clearly and communicate with the team (you do not need to be bilingual).
- Some experience with annotation, labeling, transcription, proofreading, linguistics, or other detail-oriented review work.
- A sharp, patient ear and the discipline to stay accurate over long listening sessions.
- Reliable availability of about 20 hours per week on a part-time basis.

## Nice to Have

- Prior experience on AI training-data, model-evaluation, or audio/QA annotation projects.
- A background in language, editing, voice, or audio production.

Audiobook QA Expert — Dutch

**Help make the next generation of AI-narrated audiobooks sound genuinely human.**

## 1. Overview

We're building a team of native Portuguese speakers who love audiobooks to help evaluate AI-narrated (text-to-speech) audiobooks in European Portuguese. You'll listen the way a real audiobook fan does — but with a careful, analytical ear — and flag the moments where the narration slips, so the underlying models can be made better.

This is remote, part-time work at roughly 20 hours per week (about 4 hours per day). You can be based anywhere, though we prefer people living in a region where European Portuguese is natively spoken.

## 2. What You'll Do

- Listen to AI-narrated audiobooks in European Portuguese and assess how natural, accurate, and enjoyable the narration is.
- Spot and log errors in the synthesized speech — skipped, added, or mispronounced words; numbers read incorrectly; abbreviations expanded wrongly; unnatural phrasing or intonation; audio glitches and artifacts.
- Categorize each issue and mark the exact span where it occurs, using our review tooling.
- Report on overall listener satisfaction and where the narration feels natural versus where it breaks the experience.
- Work consistently and carefully across many hours of audio without losing attention to detail.

## 3. Who We're Looking For

- Native or near-native fluency in European Portuguese, and an active, genuine audiobook listener — someone who listens regularly and has real opinions about what makes a good listen.
- Enough working English to report clearly and communicate with the team (you do not need to be bilingual).
- Some experience with annotation, labeling, transcription, proofreading, linguistics, or other detail-oriented review work.
- A sharp, patient ear and the discipline to stay accurate over long listening sessions.
- Reliable availability of about 20 hours per week on a part-time basis.

## Nice to Have

- Prior experience on AI training-data, model-evaluation, or audio/QA annotation projects.
- A background in language, editing, voice, or audio production.

Audiobook QA Expert — Portuguese (Portugal)

**Help make the next generation of AI-narrated audiobooks sound genuinely human.**

## 1. Overview

We're building a team of native Portuguese speakers who love audiobooks to help evaluate AI-narrated (text-to-speech) audiobooks in Brazilian Portuguese. You'll listen the way a real audiobook fan does — but with a careful, analytical ear — and flag the moments where the narration slips, so the underlying models can be made better.

This is remote, part-time work at roughly 20 hours per week (about 4 hours per day). You can be based anywhere, though we prefer people living in a region where Brazilian Portuguese is natively spoken.

## 2. What You'll Do

- Listen to AI-narrated audiobooks in Brazilian Portuguese and assess how natural, accurate, and enjoyable the narration is.
- Spot and log errors in the synthesized speech — skipped, added, or mispronounced words; numbers read incorrectly; abbreviations expanded wrongly; unnatural phrasing or intonation; audio glitches and artifacts.
- Categorize each issue and mark the exact span where it occurs, using our review tooling.
- Report on overall listener satisfaction and where the narration feels natural versus where it breaks the experience.
- Work consistently and carefully across many hours of audio without losing attention to detail.

## 3. Who We're Looking For

- Native or near-native fluency in Brazilian Portuguese, and an active, genuine audiobook listener — someone who listens regularly and has real opinions about what makes a good listen.
- Enough working English to report clearly and communicate with the team (you do not need to be bilingual).
- Some experience with annotation, labeling, transcription, proofreading, linguistics, or other detail-oriented review work.
- A sharp, patient ear and the discipline to stay accurate over long listening sessions.
- Reliable availability of about 20 hours per week on a part-time basis.

## Nice to Have

- Prior experience on AI training-data, model-evaluation, or audio/QA annotation projects.
- A background in language, editing, voice, or audio production.

Audiobook QA Expert — Portuguese (Brazil)

**Help make the next generation of AI-narrated audiobooks sound genuinely human.**

## 1. Overview

We're building a team of native German speakers who love audiobooks to help evaluate AI-narrated (text-to-speech) audiobooks in German (Germany). You'll listen the way a real audiobook fan does — but with a careful, analytical ear — and flag the moments where the narration slips, so the underlying models can be made better.

This is remote, part-time work at roughly 20 hours per week (about 4 hours per day). You can be based anywhere, though we prefer people living in a region where German is natively spoken.

## 2. What You'll Do

- Listen to AI-narrated audiobooks in German and assess how natural, accurate, and enjoyable the narration is.
- Spot and log errors in the synthesized speech — skipped, added, or mispronounced words; numbers read incorrectly; abbreviations expanded wrongly; unnatural phrasing or intonation; audio glitches and artifacts.
- Categorize each issue and mark the exact span where it occurs, using our review tooling.
- Report on overall listener satisfaction and where the narration feels natural versus where it breaks the experience.
- Work consistently and carefully across many hours of audio without losing attention to detail.

## 3. Who We're Looking For

- Native or near-native fluency in German (Germany), and an active, genuine audiobook listener — someone who listens regularly and has real opinions about what makes a good listen.
- Enough working English to report clearly and communicate with the team (you do not need to be bilingual).
- Some experience with annotation, labeling, transcription, proofreading, linguistics, or other detail-oriented review work.
- A sharp, patient ear and the discipline to stay accurate over long listening sessions.
- Reliable availability of about 20 hours per week on a part-time basis.

## Nice to Have

- Prior experience on AI training-data, model-evaluation, or audio/QA annotation projects.
- A background in language, editing, voice, or audio production.

Audiobook QA Expert — German

**Help make the next generation of AI-narrated audiobooks sound genuinely human.**

## 1. Overview

We're building a team of native French speakers who love audiobooks to help evaluate AI-narrated (text-to-speech) audiobooks in French (France). You'll listen the way a real audiobook fan does — but with a careful, analytical ear — and flag the moments where the narration slips, so the underlying models can be made better.

This is remote, part-time work at roughly 20 hours per week (about 4 hours per day). You can be based anywhere, though we prefer people living in a region where French is natively spoken.

## 2. What You'll Do

- Listen to AI-narrated audiobooks in French and assess how natural, accurate, and enjoyable the narration is.
- Spot and log errors in the synthesized speech — skipped, added, or mispronounced words; numbers read incorrectly; abbreviations expanded wrongly; unnatural phrasing or intonation; audio glitches and artifacts.
- Categorize each issue and mark the exact span where it occurs, using our review tooling.
- Report on overall listener satisfaction and where the narration feels natural versus where it breaks the experience.
- Work consistently and carefully across many hours of audio without losing attention to detail.

## 3. Who We're Looking For

- Native or near-native fluency in French (France), and an active, genuine audiobook listener — someone who listens regularly and has real opinions about what makes a good listen.
- Enough working English to report clearly and communicate with the team (you do not need to be bilingual).
- Some experience with annotation, labeling, transcription, proofreading, linguistics, or other detail-oriented review work.
- A sharp, patient ear and the discipline to stay accurate over long listening sessions.
- Reliable availability of about 20 hours per week on a part-time basis.

## Nice to Have

- Prior experience on AI training-data, model-evaluation, or audio/QA annotation projects.
- A background in language, editing, voice, or audio production.

Audiobook QA Expert — French

**Help make the next generation of AI-narrated audiobooks sound genuinely human.**

## 1. Overview

We're building a team of native Spanish speakers who love audiobooks to help evaluate AI-narrated (text-to-speech) audiobooks in US / North American Spanish. You'll listen the way a real audiobook fan does — but with a careful, analytical ear — and flag the moments where the narration slips, so the underlying models can be made better.

This is remote, part-time work at roughly 20 hours per week (about 4 hours per day). You can be based anywhere, though we prefer people living in a region where Spanish is natively spoken.

## 2. What You'll Do

- Listen to AI-narrated audiobooks in Spanish and assess how natural, accurate, and enjoyable the narration is.
- Spot and log errors in the synthesized speech — skipped, added, or mispronounced words; numbers read incorrectly; abbreviations expanded wrongly; unnatural phrasing or intonation; audio glitches and artifacts.
- Categorize each issue and mark the exact span where it occurs, using our review tooling.
- Report on overall listener satisfaction and where the narration feels natural versus where it breaks the experience.
- Work consistently and carefully across many hours of audio without losing attention to detail.

## 3. Who We're Looking For

- Native or near-native fluency in Spanish (US / North American Spanish), and an active, genuine audiobook listener — someone who listens regularly and has real opinions about what makes a good listen.
- Enough working English to report clearly and communicate with the team (you do not need to be bilingual).
- Some experience with annotation, labeling, transcription, proofreading, linguistics, or other detail-oriented review work.
- A sharp, patient ear and the discipline to stay accurate over long listening sessions.
- Reliable availability of about 20 hours per week on a part-time basis.

## Nice to Have

- Prior experience on AI training-data, model-evaluation, or audio/QA annotation projects.
- A background in language, editing, voice, or audio production.

Audiobook QA Expert — Spanish (US)

**Help make the next generation of AI-narrated audiobooks sound genuinely human.**

## 1. Overview

We're building a team of native English speakers who love audiobooks to help evaluate AI-narrated (text-to-speech) audiobooks in US / North American English. You'll listen the way a real audiobook fan does — but with a careful, analytical ear — and flag the moments where the narration slips, so the underlying models can be made better.

This is remote, part-time work at roughly 20 hours per week (about 4 hours per day). You can be based anywhere, though we prefer people living in a region where English is natively spoken.

## 2. What You'll Do

- Listen to AI-narrated audiobooks in English and assess how natural, accurate, and enjoyable the narration is.
- Spot and log errors in the synthesized speech — skipped, added, or mispronounced words; numbers read incorrectly; abbreviations expanded wrongly; unnatural phrasing or intonation; audio glitches and artifacts.
- Categorize each issue and mark the exact span where it occurs, using our review tooling.
- Report on overall listener satisfaction and where the narration feels natural versus where it breaks the experience.
- Work consistently and carefully across many hours of audio without losing attention to detail.

## 3. Who We're Looking For

- Native or near-native fluency in English (US / North American English), and an active, genuine audiobook listener — someone who listens regularly and has real opinions about what makes a good listen.
- Enough working English to report clearly and communicate with the team.
- Some experience with annotation, labeling, transcription, proofreading, linguistics, or other detail-oriented review work.
- A sharp, patient ear and the discipline to stay accurate over long listening sessions.
- Reliable availability of about 20 hours per week on a part-time basis.

## Nice to Have

- Prior experience on AI training-data, model-evaluation, or audio/QA annotation projects.
- A background in language, editing, voice, or audio production.

Audiobook QA Expert — English (US)

## Help map public defenders' work

We are recruiting 100 current or former public defenders for a remote research study through Project Mercator. We want to understand the work you personally did, not an ideal job description.

## What you will do

In the first survey, describe your primary responsibilities, tasks, smaller pieces of work, and which tasks need outputs from other tasks. You can also describe different versions of the same task. Written instructions and help from the study coordinator are provided. No outside research or real work product is required.

Wave 1 pays $100 per hour of reviewed survey time, up to 90 minutes ($150 maximum). If you need more time, contact the coordinator before continuing beyond the cap.

A later survey will ask selected participants to review a combined task tree and add details. Its scope and compensation will be confirmed separately before you begin.

## Who should apply

- Current or former public defenders with hands-on experience representing indigent clients in the United States.
    
- Able to describe the work you personally performed and the jurisdiction and period in which you practiced.
    
- Prior Mercor project experience is welcome but not required.
    

Experience solely as a prosecutor, law student, legal-content annotator, or lawyer without public-defense experience is not sufficient for this study. If your work was court-appointed or contract public defense, explain that arrangement so we can review the fit.

## How to participate

Apply with your resume and a short description of your public-defense experience and availability. We review applications before sending personal survey links. Applying does not automatically enroll you or start the paid survey.

Use your own professional experience, not AI-generated answers. Do not include client names, case numbers, identifying facts, confidential information, or privileged material. We do not need any documents from real cases.

Public Defenders — Paid Research Study

# **Overview**

We're looking for professionals with hands-on, real-world experience in specialized applications.

## Application Expertise and Access

You must already be running **Linux Ubuntu 24.04** as your own day-to-day desktop operating system, and be able to work in it today. **We do not provide software licenses, virtual machines, or hardware.**

**Operating Systems**

\- Linux Ubuntu 24.04 — as your primary desktop OS

\---

### About These Applications

**Linux Ubuntu 24.04** | The Ubuntu desktop itself — the GNOME environment, Files, Settings, the terminal and package management | Linux desktop users, sysadmins, DevOps and open-source developers |

## **Technology Requirement**

Must be able to capture screenshots with more than 2.5 megapixels. Satisfy this in **either** of two ways.

**Option A — a single high monitor.** Any of these aspect ratios, at or above the resolution shown:

4:3 | 1826 × 1370 | 2048×1536

16:9 | 2109 × 1186 | 2560×1440 (QHD) · 3840×2160 (4K) · 5120×2880 (5K)

16:10 | 2000 × 1250 | 2240×1400 · 2560×1600 · 2880×1800

3:2 | 1937 × 1291 | 2160×1440 · 2256×1504 · 2736×1824 · 3000×2000

21:9 (ultrawide) | 2416 × 1036 | 2560×1080 · 3440×1440 · 3840×1600

MacBook Pro 14" (3024×1964) and 16" (3456×2234) built-in displays qualify.

**Option B — a dual-monitor setup.** Two displays side by side, captured as a single spanning image. Two 1920×1080+ monitors qualify. A single 1080p screen out of that pair does not.

Application Users - Linux Ubuntu 24.04 Expert - Professionals

# **Overview**

We're looking for professionals with hands-on, real-world experience in specialized applications.

## Application Expertise and Access

You must already be running **macOS Sonoma 14.5** as your own day-to-day desktop operating system, and be able to work in it today. **We do not provide software licenses, virtual machines, or hardware.**

**Operating Systems**

\- macOS Sonoma 14.5 — as your primary desktop OS

\---

### About These Applications

**macOS Sonoma 14.5** | The macOS desktop itself — Finder, System Settings, window and Space management and the bundled applications | Mac power users, Apple-platform administrators, creative and technical professionals |

## **Technology Requirement**

Must be able to capture screenshots with more than 2.5 megapixels. Satisfy this in **either** of two ways.

**Option A — a single high monitor.** Any of these aspect ratios, at or above the resolution shown:

4:3 | 1826 × 1370 | 2048×1536

16:9 | 2109 × 1186 | 2560×1440 (QHD) · 3840×2160 (4K) · 5120×2880 (5K)

16:10 | 2000 × 1250 | 2240×1400 · 2560×1600 · 2880×1800

3:2 | 1937 × 1291 | 2160×1440 · 2256×1504 · 2736×1824 · 3000×2000

21:9 (ultrawide) | 2416 × 1036 | 2560×1080 · 3440×1440 · 3840×1600

MacBook Pro 14" (3024×1964) and 16" (3456×2234) built-in displays qualify.

**Option B — a dual-monitor setup.** Two displays side by side, captured as a single spanning image. Two 1920×1080+ monitors qualify. A single 1080p screen out of that pair does not.

Application Users - macOS Sonoma 14.5 Expert - Professionals

Explore remote opportunities with top companies worldwide. Find roles that match your skills and work preferences.

Explore opportunities

B2B Sales Expert (3+ YOE, US Only)

Quality Analyst

Medicare Advantage Members (Devoted Health) – Insight Study