6 min read · AITasks.live guides
AI Training Jobs for Experts: Doctors, Lawyers, Coders and More
By AITasks.live editorial
Published Updated Sources reviewed
Short answer: AI labs pay professionals to write hard questions in their field, grade model answers against professional standards, and write expert-quality reference answers. Published rates for expert roles are mostly in the tens of dollars to a few hundred dollars an hour, depending on the field, credential and project. The work is contract-based and project-driven, and it can stop with little notice. A credential gets you considered. It doesn't guarantee hours.
What "expert AI training" actually involves
The tasks look similar across platforms:
- Writing prompts that a strong model gets wrong in your field
- Evaluating responses for accuracy, safety and professional quality
- Writing "golden" responses, the answer a top professional would give
- Building rubrics and scenarios, such as clinical cases or legal fact patterns
Handshake AI describes its Fellows' work as "creating prompts in the Fellow's field of expertise," "evaluating or comparing AI-generated responses" and "correcting and improving content with academic citations" (Handshake AI FAQ). A closed Mercor coding listing illustrates the work: it describes crafting "domain-relevant prompts" and creating "golden responses" (Mercor — closed example).
Published pay ranges by field
These examples were checked against the original pages on 16 October 2026. Published rates are not average earnings, and eligibility varies. An advertisement or talent network does not establish that paid work is available now.
| Field | Advertised example | Important limitation |
|---|---|---|
| Software engineering | $70–$80/hr in a Mercor Coding Expert page | Closed listing, retained only as a historical example |
| Investment banking | Past rates of $100–$130/hr in Mercor's banking network | Standing talent network, not an active project or guaranteed rate |
| Management consulting | $50–$180/hr on Mercor's consulting page | Platform marketing range; actual terms are project-specific |
| Law | Up to $75/hr on Outlier's Singapore law page | Singapore eligibility; ceiling rather than a guaranteed offer |
| Machine learning / AI | Up to $150/hr on Outlier's ML page | Depends on assessment, location and project need |
| STEM and licensed professions | Published starting bands of $40–$125+/hr in DataAnnotation's FAQ | Other FAQ answers quote different bands; confirm the actual offer |
| Graduate-level expertise | $30–$125/hr in Handshake's FAQ | Education, expertise and project determine terms; some work uses per-task pay |
Alignerr's backend page gives conflicting pay ranges, so we do not use it as a rate benchmark. micro1's rates should be checked on the individual role; a third-party median is not an offer.
For a fuller picture of pay, including the low end, see How much do AI training jobs really pay?
Field-by-field: what platforms ask for
Doctors, nurses and clinicians
Listings commonly require an active license and clinical experience, and many are restricted to the US (eligibility is stated on individual roles in our Mercor directory). Typical tasks include clinical scenario writing, documentation review and grading model answers against current guidelines. Tip: list specialty, board certification and years in practice at the very top of your resume.
Lawyers
Outlier's Singapore law listing requires a JD or equivalent plus practicing experience (Outlier). DataAnnotation's professional track requires credentials in law, finance or medicine (DataAnnotation). Jurisdiction matters, so state where you're admitted.
Software engineers and ML specialists
Coding work can involve both writing and reviewing code. Expect live coding or code-review tasks: micro1 includes a coding challenge for technical roles (micro1 interview guide), and DataAnnotation mentions "LeetCode-style problems" (DataAnnotation). Listings ask for years of experience (e.g., "2+ years" or a master's in CS on Mercor — closed example; "3+ years" on Outlier).
Finance, consulting and business
Mercor recruits investment bankers, consultants and finance or Excel specialists. Its consulting page says "most engagements run 10–20 hours per week" (Mercor).
PhDs, researchers and grad students
Handshake AI says "most roles require graduate-level expertise (Master's, PhD, or Postdoc)," and projects run "from 2 weeks to 3+ months" at 5–20 hours a week (Handshake). DataAnnotation's STEM track asks for a master's or PhD, or a bachelor's plus 10+ years in the field.
Which platform fits which expert?
| If you are… | Platforms to look at | Why |
|---|---|---|
| Licensed physician / clinician | Mercor, Outlier, DataAnnotation (professional track) | Dedicated medical listings and tracks |
| Lawyer | Outlier, Mercor, DataAnnotation | Law-specific listings and tracks |
| Software / ML engineer | Mercor, Alignerr, micro1, Outlier, DataAnnotation | Coding is offered on all of them |
| Grad student / PhD / postdoc | Handshake AI, DataAnnotation (STEM), Outlier | Programs built around academic expertise |
| Finance / consulting | Mercor | Published finance and consulting roles |
This is a comparison of published specialties, not a recommendation or confirmation of openings. Our platform directory currently lists Mercor and micro1 only.
The catches experts should know
- It's contract work. Mercor, Outlier and Handshake all engage you as an independent contractor. You handle your own taxes, and there are no benefits.
- Projects end. Mercor says contributors are not automatically moved to a new project when one ends (Mercor support).
- Check the pay unit. A price per accepted task is not an hourly rate. Time spent revising or waiting can reduce the effective hourly return.
- Visa and location limits. Some Mercor listings state they can't support H-1B or STEM OPT candidates (Mercor listing). Many medical roles are US-only.
- Check your employer contract. Mercor listings say the work won't involve your employer's confidential information. Even so, read your employment agreement for outside-work or conflict-of-interest clauses before you sign up.
FAQ
Are high advertised rates guaranteed?
No. A published ceiling, past project rate or talent-network example is not a personal offer. For medical work, check the specific credential, jurisdiction, location and project requirements before comparing pay.
Do I need a license or degree for expert AI training jobs?
For expert tracks, usually yes. DataAnnotation requires licensed credentials for law, finance and medicine work. Outlier's law listing requires a JD plus practice experience. Handshake AI mostly recruits master's, PhD and postdoc candidates.
How many hours a week do expert AI training projects need?
It varies by project. Handshake says Fellows typically work 5–20 hours a week, and Mercor's consulting page says most engagements run 10–20 hours. Some listings set minimums, such as 10 hours a week.
Is expert AI training work steady?
Not reliably. Projects are temporary and can be extended, shortened or ended early. Plan for it as variable side income.
Which platform is best for experts?
No single platform is best. It depends on your field and location. Mercor publishes many high-rate specialist roles, Handshake AI focuses on graduate-level academics, and DataAnnotation and Outlier have dedicated expert tracks. Compare roles that fit your background; several applications still do not guarantee steady work.
See expert roles open right now: Live AI training jobs · Mercor expert listings · All platforms
Before you apply: Write a resume that gets accepted · Pass the assessments
Disclosure: AITasks.live may earn a referral fee if you join a platform through our links. Platforms cannot pay for placement. How we make money.
Find your next AI training opportunity
Explore Mercor and micro1 roles that match your skills and location. Acceptance and project availability vary.
We may earn a referral fee if you join through these links.