Research-level work for AI
AI Training Jobs for PhDs
Measured from 61 distinct advertised rates, across 7 platforms. 12 of the 75 open roles here state no hourly rate; 2 advertise only a maximum ("up to $X/hr"), which is excluded.
AI labs pay people with a doctorate or a master's by the hour to write research-level problems in their field, find where a model's reasoning breaks and write the answer an expert would give. 75 roles for PhDs are open right now on 10 platforms, including Alignerr, micro1 and Mercor. Every role is checked against its platform's job board daily.
Live roles for PhDs
Updated 4 min agoShowing the 1 roles whose requirements name no degree, every country. See every role
Some links on AITraining.jobs are referral links. If you sign up through one, AITraining.jobs may earn a commission at no extra cost to you. It never changes what a platform pays you.
PhDs and master's graduates in any field
1 openRoles open to graduate degree holders across disciplines or across STEM: write hard questions in your own field and grade a model's answers against them.
STEM Expert (MSc./PhD) (opens the job details)
micro1 · Math & STEM · Remote· 1d agoDetails
$60–$120/hr
Before you apply
This is training and evaluation work for AI models, not a research position: you will not publish or keep rights to what you write. If you hold a university or lab appointment, check its policy on outside work first, and never bring unpublished or confidential research into a task. A legitimate platform never charges you to apply.
Pay by specialty
Each specialty's typical pay is measured over its own open roles the way every band on this site is (the 25th percentile of what roles pay (the middle of each advertised range) to the 90th-percentile advertised ceiling), and shown only where at least three distinct rates are posted.
What the roles ask for
Read from the requirements of 35 of the 75 open roles for PhDs: the lines each role's pop-up lists from its posting. A role counts once, at the lowest degree it accepts.
The roles that name experience ask for a median of 2 years.
What platforms pay to find you
micro1, Turing and AfterQuery post a flat referral bonus on 28 of the 75 open roles for PhDs: what the platform pays whoever refers the person it accepts, a measure of how hard these people are to find. The platform pays it, and it never comes out of your pay.
Who hires PhDs for AI work
| Platform | Open roles | Pay stated |
|---|---|---|
 Alignerr | 25 | $20–$120/hr |
| micro1 | 18 | $45–$200/hr |
 Mercor | 10 | $50–$110/hr |
 Turing | 9 | Not stated |
 Braintrust | 3 | $30–$150/hr |
 Terac | 3 | Not stated |
 AfterQuery | 2 | $100/hr |
 Handshake AI | 2 | up to $75/hr |
 Welo Data | 2 | $50–$70/hr |
 Anyone AI | 1 | $150/hr |
How each platform checks you
The platforms hiring PhDs now, in the words of each one's onboarding guide, with the source of every line. The full steps are on each platform's page.
- Assessment
- Role assessments and the Zara AI interview. How to apply
- Interview
- AI interview (recorded, scored by AI)
- Identity check
- Once you are matched to a project: a driver's license, state ID, work permit or permanent residence card from an accepted country, photographed live through Persona, plus a recent selfie. Source
- Assessment
- The Zara AI interview (plus a role-specific skill check). How to apply
- Interview
- AI interview (recorded, scored by AI)
- Identity check
- Valid government-issued photo ID (passport or driver's license) for identity verification, matching the legal name on your application. Source
- Assessment
- AI Video Interview (plus role-specific Form assessments). How to apply
- Interview
- AI interview (recorded, scored by AI)
- Assessment
- Role-specific Turing vetting (proctored tests, AI interview, human interviews). How to apply
- Interview
- AI interview (recorded, scored by AI)
- Identity check
- A valid, unexpired, physical government photo ID: a passport or driver's license (Nigerian users may also use a physical NIN card). Source
- Credentials
- It describes three stages: screening and matching on your profile (skills, domain, credentials, availability), automated testing (coding challenges in 18-plus languages, RLHF and SFT-style task evaluations, and an adaptive AI interviewer), and expert human review with deep-dive interviews for complex or high-stakes roles. Source
- Assessment
- Braintrust screening interview (AIR / AI skills interview). How to apply
- Interview
- AI interview (recorded, scored by AI)
- Identity check
- Government-issued photo ID and ability to take a selfie for Identity Verification (verification is run through Persona; images are not stored or displayed on Braintrust). Source
- Assessment
- Per-opportunity screening (often a short AI voice interview), plus an optional onboarding interview. How to apply
- Interview
- AI interview (recorded, scored by AI)
- Identity check
- A government-issued photo ID before your first cash-out. Source
- Assessment
- AfterQuery pool skills assessment (plus an interview step). How to apply
- Interview
- Video interview
- Identity check
- For verification and payouts: a government-issued photo ID, a device with a camera for a selfie, a phone that can receive a text code, a bank account in your own legal name in the same country you select in Stripe, and your SSN if you are in the US. Source
- Credentials
- AfterQuery says it welcomes 'experts with advanced degrees, significant industry experience, or demonstrated expertise in their field,' and that every expert goes through a verification process that includes credential checks and expertise validation. Source
- Assessment
- Project assessment (per project, after an invitation). How to apply
- Interview
- None published
- Identity check
- A government photo ID (driver's license, state ID, passport or green card) and a phone or tablet with a camera for the Persona ID and selfie check. Source
- Assessment
- Timed Welo Works assessments, then a client exam (per role). How to apply
- Interview
- None published
- Assessment
- Interview (format not published), then project onboarding. How to apply
- Interview
- Interview
Where these roles are posted
Roles a platform posts for one country, and roles whose title names a language. Each place opens its own page; each count opens these roles.
A licensed or credentialed professional?
One profile for every AI training platform. Companies ask us for people with a specific language, country or credential: add yours once, and when a role fits, we introduce you, only with your yes. Free.
Get new AI training jobs by email
The newest paid roles across every platform we track, with real hourly pay, in one weekly email. Free, and unsubscribe anytime.
We email new roles, not spam. One click to unsubscribe.
Frequently asked questions
How much do AI training jobs for PhDs pay?
Across the 61 distinct hourly rates posted for roles for PhDs right now, typical pay runs from $80 to $200 an hour, with a median of $80. The highest posted rate is $200 an hour, at micro1. 12 of the 75 open roles for PhDs state no hourly rate, so check each listing.
Do I need a finished PhD?
Not always. Many roles take a master's or a PhD in progress, and several accept postdoctoral researchers or equivalent research experience; the best-paid ask for a completed doctorate in a named field. Each listing states the degree, the field and the countries it accepts, and platforms ask for proof of the degree when they verify you. Of the 35 open roles for PhDs whose requirements we read, 26 ask for a PhD, 5 a master's, 3 a bachelor's or a degree and 1 name no degree.
What does a PhD do in AI training?
You write problems from your own research area that the best models get wrong, with a verified solution, grade a model's step-by-step reasoning against it, and explain where it went wrong. Some projects add rubric writing or reviewing other experts' problems. It is remote, hourly contract work, and the hours depend on the project.
How do platforms check PhDs?
Most platforms start with an assessment in your field before any paid work, and each listing says what it asks for. Of the platforms hiring PhDs now, Alignerr, micro1, Mercor, Turing and Braintrust run an AI interview (a recorded interview scored by AI) for many roles. Alignerr, micro1, Turing, Braintrust and Terac check a government photo ID, usually with a selfie, before you are paid. Turing and AfterQuery say they check credentials such as degrees. The section on how each platform checks you quotes each guide with its source.
More work by profession
AI training jobs for mathematicians
Math experts, PhD mathematicians, Lean 4 and statistics.
AI training jobs for physicists
Research physicists, astronomers and physics experts.
AI training jobs for chemists
Synthetic, analytical, computational chemists and biochemists.
AI training jobs for biologists
Bioinformatics, molecular biology, neuroscience and life sciences.
AI training jobs for environmental scientists
Environmental, earth, climate and agricultural scientists.
AI training jobs for nuclear and CBRN experts
Nuclear, chemical, biological, explosives and defense experts.
AI training jobs by profession
Every profession with open roles, with live counts and pay.
Math and STEM
Every math and science role, with live pay by platform.
Expert fields in demand
Which professions AI training roles ask for, counted from their titles.