experts / 03 · ai training gigs
AI training gigs. Specialist work, specialist rates, shown first.
Twenty-seven open, remote, flexible gigs for people with real credentials: physicians, attorneys, PhDs, licensed engineers, accountants, translators, and trades. Each one below states who qualifies, what the work is, the weekly hours, and the hourly range, so you know the deal before you apply.
Hero image · 840 × 1050
open gigs
27
across seven groups, all remote
lowest hourly floor
$40/hr
no listing on the board pays less
top of range
$300/hr
board-certified radiology
reply to applications
1business day
read by a person on the data team
What are AI training gigs?
AI training gigs are short, paid, remote engagements in which someone with real expertise produces or checks the data a model learns from. A litigator writes a fact pattern and grades the model's answer. A radiologist marks what the model missed on a scan. A plumber reads an AI repair walkthrough and says whether it would pass inspection. The AI teams who buy this data need it from practitioners, because a model trained on the open internet already knows what the internet knows and nothing more.
eeko runs these gigs through its private expert network, not an open crowd. Each listing below is a real project type with a stated hourly range, a weekly commitment that fits around a practice or a job, and a credential we verify before any paid work. Approved work is reviewed by a second expert, delivered to the buyer with a signed provenance record on every item, and paid at the rate on the listing. Hands-on occupations have their own capture programs, listed separately.
Gig board image · 2100 × 900
gigs / legal
Legal AI training gigs for practicing attorneys.
Writing, redlining, and grading work for lawyers with an active bar admission. The models being trained will draft and analyze for other lawyers, so the bar is what a partner would sign.
legal / 01OPEN
Litigator: hard cases, honest grades
A gig for courtroom lawyers. The buyers are training models to reason about disputes, and they need practitioners to produce the difficult scenarios and score the results. You invent fact patterns in your own practice area and write the answer you would stand behind, rate how well a model handled the governing statutes, precedent, and procedural posture, and pinpoint the step where its argument falls apart in front of a judge. Some projects put two model-written briefs side by side and ask which is stronger, and why.
- Who: admitted attorneys with at least five years of litigation behind them, including genuine motion work and time in trial or arbitration; appellate, antitrust, commercial, employment, IP, securities, and white-collar backgrounds all wanted
- Pay: $150 to $250 per approved hour
- Hours: 5 to 15 a week, remote
legal / 02OPEN
Transactional attorney: how deals are really papered
A gig for deal lawyers. Models are being taught how transactions get documented, and only someone who has closed them can show it. You read agreements a model drafted, from acquisition and credit facilities to prospectus and closing deliverables, decide what would never survive a negotiation, note what is absent or unenforceable, and record how the other side would respond. The output is an annotated library of clauses and positions with your reasoning attached.
- Who: bar-admitted lawyers with five or more years of private equity, leveraged finance, capital markets, or M&A work at a large firm or in-house, comfortable in Delaware, New York, or English law forms
- Pay: $175 to $275 per approved hour
- Hours: 5 to 15 a week, remote
legal / 03OPEN
Patent practitioner: claims, prior art, office actions
A gig for registered patent practitioners. Starting from an invention disclosure, you draft the claims. Given a model's claim set, you pull it apart for scope and written support. Given prior art, you decide what is novel and what is obvious. Given a model's reply to an examiner, you judge whether it would move the application forward. Work is sorted by technical discipline so you stay inside your own, whether that is software, electrical, biotech, mechanical, or chemical.
- Who: attorneys and agents registered with the USPTO, EPO, or UKIPO who hold a technical degree and have several years of prosecution experience; litigation or licensing work on top is welcome
- Pay: $175 to $275 per approved hour
- Hours: 5 to 15 a week, remote
gigs / finance
Finance, tax, and actuarial gigs for credentialed professionals.
Problem design and grading for people who hold a CPA, CFA, fellowship, or a quantitative doctorate. The errors that matter here are the ones a desk, a committee, or an auditor would catch.
finance / 04OPEN
Tax specialist: corporate and cross-border questions
A gig for corporate and international tax professionals. You create realistic structuring and compliance scenarios and solve them in full, then turn to a model's attempts at similar questions and mark them against the Code, the regulations, and the applicable treaty. Topics run from corporate and partnership taxation through withholding on cross-border payments, transfer pricing, treaty claims, and state and local issues. The question underneath every task is whether following the model's advice would end in a penalty or a lost audit.
- Who: a CPA, a JD, or a tax LLM plus at least five years of corporate or international tax practice at a large firm, a Big Four practice, or in-house, with a working grasp of the Internal Revenue Code and Treasury regulations; OECD guideline experience helps
- Pay: $150 to $225 per approved hour
- Hours: 5 to 15 a week, remote
finance / 05OPEN
Quantitative finance PhD: pricing, risk, and markets
A gig for quantitative researchers who have sat on a desk or in a risk function. You write research-grade problems in stochastic calculus, option pricing and hedging, portfolio construction, risk measurement, and the mechanics of markets, and you solve them. Then you score a model's attempts line by line and audit the quantitative analysis and code it produced, looking for exactly the errors a trading desk or a risk committee would refuse to accept.
- Who: a doctorate in statistics, economics, physics, mathematics, or financial engineering, time at a quant fund, on a trading desk, or in risk, and fluency in Python or C++ as well as the mathematics underneath the models
- Pay: $175 to $275 per approved hour
- Hours: 5 to 15 a week, remote
finance / 06OPEN
CFA charterholder: research the committee would accept
A gig for buy-side and sell-side professionals. Models are producing equity research, discounted cash flow and comparables valuations, credit write-ups, and investment memos. Your job is to read them the way an investment committee would: mark factual errors, flawed method, and poor judgment, rewrite the weak sections to a standard you would present yourself, and construct worked valuation cases from public filings for the model to learn from.
- Who: CFA charterholders or the equivalent with at least five years in equity or credit research, running portfolios, or corporate finance, with strong accounting and modeling and the ability to defend a view in writing
- Pay: $150 to $225 per approved hour
- Hours: 5 to 15 a week, remote
finance / 07OPEN
Fellow actuary: what a signing actuary would strike out
A gig for fellowship-level actuaries. You design reserving, pricing, capital, and experience-study problems and solve them in full, score a model's answers against the standards of practice, and audit model-generated actuarial analysis for the kind of assumption or method you would never put your name to. Life, health, pensions, and property and casualty practice areas are all needed.
- Who: FSA, FCAS, FIA, or an equivalent fellowship, with professional practice in your area
- Pay: $150 to $225 per approved hour
- Hours: 5 to 15 a week, remote
gigs / medicine
Medical AI training gigs for licensed clinicians.
Case authoring, imaging annotation, and safety grading for board-certified physicians, pharmacists, and mental health clinicians. The question behind every task is whether a patient would be harmed.
medicine / 08OPEN
Radiologist: what the model missed on the scan
A gig for diagnostic radiologists who can read from home. You annotate findings on de-identified studies across modalities, write the reference report for each, and then score a model's interpretation of the same study, looking for the missed finding, the overcall, and the reporting error that would have changed what the treating team did next. Work is divided by subspecialty so you read within your own: neuro, body, chest, musculoskeletal, breast, or emergency.
- Who: MD or DO with board certification in diagnostic radiology and an active license; fellowship training is a plus, and you need a home workstation adequate for diagnostic reading
- Pay: $200 to $300 per approved hour
- Hours: 5 to 15 a week, remote
medicine / 09OPEN
Physician: clinical reasoning in your specialty
A gig for board-certified specialists. The tasks are the ones you do on rounds, written down: compose a clinical case with the reasoning you would teach, score the differential, workup, and plan a model proposed against current guidelines and evidence, say plainly which model answers would harm the patient, and fix the model's attempts at talking to patients. Specialties open now include cardiology, emergency medicine, internal medicine, obstetrics, oncology, pediatrics, psychiatry, and surgery.
- Who: MD or DO, board-certified, actively licensed, recently in clinical practice, and able to write clearly about evidence and about uncertainty
- Pay: $175 to $275 per approved hour
- Hours: 5 to 15 a week, remote
medicine / 10OPEN
Clinical pharmacist or pharmacologist: dosing and interactions
A gig for pharmacists and pharmacologists. Models answer questions about dosing, adjustments for kidney and liver function, interactions between drugs, pharmacokinetics, and adverse effects. You decide whether those answers match the product label and the primary literature, you flag anything a patient should never be told, and you write reference cases the model can be trained on.
- Who: PharmD with a board certification such as BCPS, BCOP, or BCCCP, or a clinical pharmacology PhD, backed by hospital or pharmaceutical industry experience
- Pay: $100 to $160 per approved hour
- Hours: 5 to 15 a week, remote
medicine / 11OPEN
Psychologist or psychiatrist: the reply a clinician would give
A gig for licensed mental health clinicians. Models hold conversations with people who describe anxiety, depression, trauma, suicidal thinking, and relationship distress. You read those exchanges and score them on safety, on the quality of the formulation, and on whether the response was therapeutic, then write the reply a trained clinician would have given. Some projects also ask you to build structured assessment cases.
- Who: a licensed PhD or PsyD in clinical psychology, or an MD with board certification in psychiatry, currently practicing; training in CBT, DBT, or psychodynamic approaches is useful
- Pay: $150 to $225 per approved hour
- Hours: 5 to 15 a week, remote
gigs / science_and_research
Science and research gigs for PhDs.
New problems, rigorous solutions, and step-level verification for reasoning models. Public problem sets are already in the training data; what buyers need is material a model has never seen.
science / 12OPEN
Mathematician: new problems, checked proofs
A gig for PhD mathematicians. You compose problems from qualifying-exam level upward and write rigorous solutions, check a model's proofs one step at a time and name the exact line where the argument breaks, build counterexamples, and compare rival proofs for correctness and elegance. Demand is strongest in number theory, combinatorics, analysis, algebra, topology, probability, PDEs, and optimization.
- Who: a doctorate in mathematics or a neighboring field; postdoctoral or faculty experience helps, and an olympiad background is welcome though not required
- Pay: $125 to $200 per approved hour
- Hours: 5 to 20 a week, remote
science / 13OPEN
Physicist or chemist: derivations and paper checks
A gig for physical scientists with a publication record. You set graduate-level problems and write the full derivation, score a model's attempts for mistakes in reasoning, units, and approximation, read a model's summary of a research paper next to the paper itself and say what it got wrong, and judge whether a proposed experiment would actually test the hypothesis. Theorists and experimentalists are both needed.
- Who: a doctorate in chemistry, physics, chemical engineering, or materials science, plus published work
- Pay: $125 to $200 per approved hour
- Hours: 5 to 20 a week, remote
science / 14OPEN
Biologist or bioinformatician: pipelines and experimental design
A gig for life scientists, bench or computational. You score a model's reasoning in cell and molecular biology, immunology, genomics, and protein science, audit the pipelines and code it writes for sequencing data, variant calls, single-cell experiments, and structure prediction, point out the confounds and missing controls in experiments it proposes, and write the reference answers it should have produced.
- Who: a doctorate in computational biology, bioinformatics, genetics, biochemistry, or biology, with publications and hands-on practice in the lab or at the keyboard
- Pay: $125 to $200 per approved hour
- Hours: 5 to 20 a week, remote
science / 15OPEN
Economist: the referee's read
A gig for PhD economists. You write graduate-level problems in econometrics, macro, and micro, each with a worked solution, score a model's answers, and review the empirical analysis, identification strategies, and policy conclusions a model generates with the eye of a journal referee deciding whether to send the paper back.
- Who: a doctorate in economics or a nearby quantitative discipline, with research experience in academia, a central bank, a consultancy, or industry
- Pay: $125 to $200 per approved hour
- Hours: 5 to 15 a week, remote
gigs / engineering_and_software
Engineering and software gigs for licensed and senior practitioners.
Evaluation design, code compliance, and CAD review. The standards are the ones you already work under, and the buyers want to know where a model would fail them.
engineering / 16OPEN
ML researcher: evaluations that cannot be gamed
A gig for published machine learning researchers. You design evaluation tasks that separate real capability from memorized patterns, write the adversarial and edge-case prompts your specialism makes you good at, audit research code, experiments, and claimed results that a model produced, and score its reasoning on problems in algorithms, systems, and theory.
- Who: a doctorate in statistics, computer science, or a neighboring field with publications in ML, NLP, systems, or theory, strong engineering ability, and ideally experience training or evaluating large models
- Pay: $150 to $250 per approved hour
- Hours: 5 to 20 a week, remote
engineering / 17OPEN
Licensed structural or civil engineer: code compliance
A gig for professional engineers who stamp drawings. You audit the load paths, member sizes, calculations, and code readings a model produced against the standards you practice under, whether that is ACI, AISC, ASCE 7, IBC, or the Eurocodes, score the model's design explanations, and write reference problems drawn from real site conditions.
- Who: a PE, CEng, or equivalent license and professional practice in geotechnical, civil, or structural engineering
- Pay: $125 to $190 per approved hour
- Hours: 5 to 15 a week, remote
engineering / 18OPEN
Mechanical or civil engineer: CAD and drawings
A gig for degreed engineers who live in a CAD or BIM tool. You inspect models, drawings, and the design explanations an AI produced for mistakes, missing tolerances, and code violations, and you label parts and assemblies so a model can learn what each component is.
- Who: an engineering degree, professional experience, and working fluency in at least one CAD or BIM package
- Pay: $50 to $90 per approved hour
- Hours: 5 to 20 a week, remote
gigs / review_and_grading
Review and grading gigs for qualified professionals.
Higher-volume evaluation work: read what a model produced, decide whether it is right, and write the version a client should have had. Broader entry requirements, still paid as professional work.
review / 19OPEN
Software engineer: judge AI-written code
A gig for engineers who ship production code. You read code an AI system wrote and decide whether it is correct, secure, and idiomatic, then write a short, exact note on the defect and your fix. Some tasks put two solutions to one problem side by side; some ask you for a reference implementation. Python, TypeScript, and Java dominate.
- Who: professional experience shipping production code, the habit of reading unfamiliar code quickly and catching subtle bugs, and clear written English
- Pay: $40 to $90 per approved hour
- Hours: 10 to 30 a week, remote
review / 20OPEN
Lawyer: the answer the client should have had
A gig for admitted lawyers in everyday practice areas. You read AI answers to the questions people ask about contracts, employment, property, consumer rights, and procedure, score them for accuracy and completeness, flag the mistakes a client might act on, and where the answer is wrong, write the one they should have received.
- Who: a law degree plus bar admission, or your jurisdiction's equivalent, with current or recent practice and the ability to explain the law in plain language
- Pay: $80 to $150 per approved hour
- Hours: 5 to 20 a week, remote
review / 21OPEN
Physician, nurse, or pharmacist: three questions per answer
A gig for licensed clinicians. You read AI answers to health questions and to clinical scenarios and settle three questions of your own: is the information right, is there a dangerous omission, and would a patient come away misled. You record a short judgment and, occasionally, rewrite the answer.
- Who: a current license as a physician, nurse practitioner, RN, or pharmacist
- Pay: $50 to $150 per approved hour, with physicians starting at $100 and nursing and pharmacy at $50
- Hours: 5 to 15 a week, remote
review / 22OPEN
Accountant or bookkeeper: the correct treatment
A gig for accountants who know what small business owners actually ask. You review AI answers on bookkeeping, payroll, small-business tax, and financial statements, identify the errors, and write the correct treatment with the reasoning behind it.
- Who: a CPA or ACCA qualification, or several years in practice
- Pay: $50 to $100 per approved hour
- Hours: 5 to 20 a week, remote
review / 23OPEN
STEM graduate: exam problems and step-by-step grades
A gig for people with graduate-level command of mathematics, physics, chemistry, or statistics. You set exam-style problems, write complete solutions, and then grade AI attempts one step at a time, noting precisely where the reasoning goes off course. Teaching or tutoring experience helps.
- Who: a graduate degree, or equivalent depth, in at least one of the four subjects
- Pay: $45 to $90 per approved hour
- Hours: 10 to 30 a week, remote
gigs / trades_and_other
Trades, property, hospitality, and language gigs for working professionals.
Not every expert has a doctorate. These gigs pay for the judgment that comes from years on the tools, in the market, on the floor, or in a language.
trades / 24OPEN
Plumber, electrician, or HVAC tech: explain the job, check the advice
A gig for licensed trades and people with years on the tools. You describe how real jobs get diagnosed and done, read AI-written repair walkthroughs and flag anything unsafe or anything that would fail inspection, and label photographs of installations and faults. Want to record your work on site instead? See the skilled trades capture program.
- Who: a trade license or several years of hands-on experience, and the ability to put what you do into plain written English
- Pay: $40 to $75 per approved hour
- Hours: 5 to 15 a week, remote
property / 25OPEN
Real estate agent or appraiser: listings and values
A gig for licensed agents and appraisers. You score AI-written listings, the comparable-sales logic behind a price, and valuation explanations against your own knowledge of the market, and flag any statement a licensed professional could not make.
- Who: an active real estate or appraisal license and several years of transactions or valuations
- Pay: $40 to $70 per approved hour
- Hours: 5 to 15 a week, remote
hospitality / 26OPEN
Hotel or restaurant GM: rate the guest conversation
A gig for people who have run a front-of-house operation. You read exchanges between a guest and an AI concierge or an AI booking assistant and score them as a good manager would: whether the guest got what they needed, whether the tone was right, and what should have happened instead.
- Who: several years managing a hotel, a restaurant, or a front-of-house team
- Pay: $40 to $65 per approved hour
- Hours: 10 to 25 a week, remote
language / 27OPEN
Translator or linguist: Spanish, Portuguese, French, Hindi, Arabic
A gig for native speakers with strong written English. You judge translations, rewrite stilted AI output so it reads the way a native speaker would write it, and transcribe short audio clips. Translation, teaching, or editing experience helps.
- Who: native fluency in Spanish, Portuguese, French, Hindi, or Arabic, plus strong written English
- Pay: $40 to $75 per approved hour
- Hours: 10 to 30 a week, remote
how_it_works
From the listing to the first paid hour.
The same four steps for every gig on this page. A person is on the other end of each one.
STEP 01
Apply
Name the gig on the expert application, or apply to the network if nothing fits yet. Five minutes. Reply within one business day.
STEP 02
Verify
We confirm your license, degree, or certification and your identity, privately, before any paid work. Every item you make will trace to a named, verified expert.
STEP 03
Calibrate
A short paid task on the project's own guidelines, so you and the domain lead agree on what good looks like before volume starts.
STEP 04
Work and get paid
Remote, your hours, the rate on the listing. Work that passes second-expert review is itemized on your statement and paid on schedule.
what_every_gig_shares
What all twenty-seven gigs have in common.
Peer review image · 1200 × 900
The terms behind the listings
- Rates in writing — the range on the card is the range on the project brief, and the brief's rate is what approved hours pay. No repricing once you are in.
- Rates that rise — harder projects and a longer record of approved work move you up the range, and sometimes above it.
- Paid calibration — any assessment task is paid at the project rate. We do not run unpaid trials.
- Peer review, with reasons — rejected work comes back with the reviewer's note. Disputed calls go to a domain lead.
- Private credentials — buyers receive a signed provenance record per item, not your name. See how every item is verified.
- No exclusivity — keep working wherever else you like. Already on another platform? Send us your rate.
Fields collage image · 2100 × 900
related
More from the eeko data practice.
faq
AI training gigs, asked before applying.
What are AI training gigs?
AI training gigs are short, paid, remote engagements in which a qualified person produces or checks the data an AI model learns from. Depending on the field that means writing problems and reference answers, grading a model's output against the real standard, labeling images or documents, reviewing code, or rating conversations. They are contract work, not employment: each gig has a defined task, an hourly range, an expected weekly commitment, and a review process, and you take only the ones that fit around your practice, job, or research.
How much do AI training gigs pay?
Every listing on this page shows its hourly range, and no listing pays less than $40 an hour. Licensed and research-level work pays substantially more: board-certified radiologists are listed at $200 to $300 per approved hour, transactional and patent attorneys at $175 to $275, and PhD scientists at $125 to $200. You are paid at the listed rate for each approved hour, and rates move upward as projects get harder and your record of approved work grows.
Are AI training gigs full-time jobs?
No. These are flexible, remote contract engagements designed to sit alongside your main work. Most listings expect between five and twenty hours a week, and a few of the review and grading gigs run up to thirty. You set your own hours within each project's deadline. There is no exclusivity, no minimum commitment across projects, and no obligation to accept the next one. Many experts in the network take a project, pause for a month, and pick up another when it suits them.
How do I apply for an AI training gig?
Use the expert application and name the gig or gigs you are interested in. A member of the eeko data team reads it and replies within one business day. If you are a fit, we verify your license, degree, or certification privately, then set a short paid calibration task built on the project's own guidelines. Once that passes, you receive the project brief with the rate, the hours, and the deadline, and you decide whether to accept. There is no automated interview and no unpaid trial.
What if there is no gig in my field right now?
Apply anyway and tell us your field and credentials. The listings here are the ones open today; requests from AI teams open new ones regularly, and when a project matches your specialty we invite you directly rather than posting it and hoping you see it. Being in the network costs nothing and commits you to nothing. We also run capture programs for hands-on occupations such as the trades, nursing, cooking, and warehouse work, which are listed separately.
Do AI training gigs require a license or degree?
Most do, and each listing says exactly what. Medical, legal, engineering, actuarial, and accounting gigs require the relevant license or professional qualification; research gigs require a PhD or equivalent published work; the trades gigs require a trade license or several years on the tools. A few review gigs accept a graduate degree plus professional experience. Whatever the requirement, we verify it before any paid work, because every item we deliver has to trace back to a qualified human.
Pick a gig. Or let the next one find you.
Apply once with your field and credentials. We verify privately, show you the rate and the scope, and you decide. A person replies within one business day. No deck. No unpaid test. No queue.
Experienced within
markets_served
Markets served.
As an enterprise AI agency, eeko systems delivers production AI systems remote-first across the United States and internationally — including these markets: