AI Training Jobs for Doctors: What Labs Pay Physicians to Review AI

You have a medical degree, a license, or a specialty, and you keep seeing AI training work advertised at rates that look too high for part-time review. We list 128 medicine and health roles on this site as of October 23, 2026. Here is what they pay, who is hiring, and what the listings actually ask for.

TL;DR

• 128 of the 1,687 open listings on the site sit in Medicine & health, as of October 23, 2026.
• The median top hourly rate is $100/hr across the 121 medicine listings that state a rate. The 10th percentile is $70/hr, the 90th is $200/hr, and the highest is $450/hr.
• micro1 (33), AfterQuery (26) and Mercor (25) hold most of the medicine listings we list.
• None of the 128 are beginner-friendly. Every listing expects a medical credential or clinical experience.
• 82 of the 128 say only "Remote", with no country named. Remote does not mean worldwide.

AI labs pay physicians to read model output, grade it against clinical standards, and write the clinical questions and reference answers that go into training data, and the median top rate for that work right now is $100/hr.

What AI labs pay doctors to do

The listing titles tell you most of what you need. We see Oncologist, Dentist, Clinical Researcher, Medical Professional, Medical Coding Specialist and Licensed Physicians, Medical Content Reviewer. One listing, Healthcare Attorney (BigLaw Firms), sits in the medicine field because it asks for legal expertise in a health context rather than clinical practice.

The work behind those titles is review and writing. You read what a model produced and judge whether it holds up: a differential that misses something, a drug interaction stated wrongly, a treatment pathway that skips a step. You write items in your own specialty and supply the answer key. Some roles are bilingual, where you check a model's medical output in a second language.

This is contract work, not employment. Projects start and end, and the hours depend on how much data the client needs that month. None of the 128 listings promise a set number of hours, and we would not treat any of these rates as a salary.

How many roles are open and what they pay

There are 128 open medicine and health AI training jobs on the site right now, out of 1,687 overall. Of those 128, 121 state an hourly rate. The other 7 say nothing about pay, and the rate is given during onboarding. Here are eight listings that show the spread.

Where a listing shows a band, that is the band it published. Where it shows one figure with "up to", treat it as a ceiling, not a promise. A $450/hr top rate on an oncology project is the top of one project's range. It does not mean every oncologist who applies will be offered that, and we have no data on what anyone was actually paid.

Which platforms hire doctors most

Nine platforms run medicine listings right now. These are the ones with the most, with what we learned from going through their screening ourselves.

micro1

33 medicine listings, the largest share. Median top rate $125/hr, with the 10th percentile at $72/hr and the 90th at $224/hr. The highest is $400/hr. Eleven listings name specific countries, mostly the United States, three say worldwide, and 19 say only "Remote". A few ask for Japanese, Cantonese, Vietnamese or Korean.

You apply on the listing page at jobs.micro1.ai and complete an AI recruiter interview. It asks about your experience and, for technical roles, tests the skill. If you pass, you wait to be matched to a project when one opens. Do that interview somewhere quiet with a working camera and mic, because it is recorded and reviewed.

Mercor

25 medicine listings, and the highest median of the group at $135/hr. The 10th percentile is $70/hr, the 90th is $199/hr, and the top listing reaches $450/hr. Three of the 25 do not state pay. Six name a country, among them the US, China, Germany and Spain.

You upload a resume and apply to one specific listing at work.mercor.com. Most ask for a short AI-led video interview about your background, and some add a domain assessment. Apply to the listing that matches your specialty most closely. Accepting you once makes the next match faster, and projects start and end.

AfterQuery

26 medicine listings, median top rate $88/hr, 10th percentile $68/hr, 90th percentile $124/hr, highest $160/hr. None of the 26 name a country or say worldwide, so every one reads as "Remote" only. Expect a screening task that looks like the real work. Listings here refresh weekly, so a role can close between refreshes.

SME Careers

13 medicine listings, median top rate $90/hr, highest $150/hr. All 13 name their eligible countries, which almost no other platform does. You fill in a profile and take a skills assessment in your subject at sme.careers. Passing puts you in the pool for that project.

The rest

Mindrift has 11 medicine listings at a $70/hr median, Ethos 6 at $100/hr, Handshake AI 6 at $88/hr, Braintrust 4 at $85/hr, and Turing 4 with no hourly rates stated. You can compare them in more detail on our platforms compared side by side page, and read how we vet platforms before you spend time on any application.

What gets you accepted

The credential the listing names is the one that matters. A title like Oncologist or Dentist is asking for that specialty, not a general medical degree. The Braintrust listing states in the title that it wants licensed physicians in the US. The Medical Coding Specialist role at micro1 asks for coding work specifically, and the Healthcare Attorney role at micro1 wants a lawyer with health sector experience, not a clinician.

Board certification shows up less often than a license, and when it does, it comes from the project rather than the platform. Read the listing text for the project you are applying to.

Every platform screens differently, so prepare for the format you are walking into rather than for a question bank:

• micro1: the AI recruiter interview is a conversation about your experience. Have your training path and any teaching or reviewing you have done ready to describe in plain terms.
• AfterQuery: the screening task is the real work. Do it carefully and show your reasoning, because a rushed answer is the usual reason a strong candidate fails a task like this.
• Mercor: the video interview rewards specificity. Name the conditions you treat or the literature you follow, and apply to the narrow listing that fits rather than the broadest one.
• SME Careers: the skills assessment is a subject test. Review your own field before you sit it, not general medicine.

We have no data on acceptance rates, how long reviews take, or why individual applicants were turned down. We only know the steps each platform publishes.

Who can apply from where

Of the 128 medicine listings, 43 name specific countries, 3 say worldwide, and 82 say only "Remote". The 82 are the ones to check carefully. A listing that says only "Remote" has not said it is open to your country, and the platform may confirm eligibility later.

The United States is named on 40 medicine listings. Germany appears on 13. India, the Philippines, Pakistan, Bangladesh, Brazil, Singapore, Malaysia, Indonesia, Thailand and Vietnam are each named on 9, mostly through SME Careers and its regional projects.

Language requirements are a smaller but real category. Japanese appears on 3 medicine listings, Cantonese and Vietnamese on 2 each, and Korean, Chinese, German and Spanish on 1 each. If you practise in a second language, those listings narrow the field for you.

Pay, tax and visa treatment depends on where you live, so check your local rules before you accept a contract.

Where to go from here

Our Medicine & health AI training jobs page carries the live list behind every figure above, filtered to this field. If you want to see what else the site covers, the all open AI training jobs page has the full 1,687. Some of the apply links on this site earn us a referral reward when someone is hired, and you can read how referral links work here. We do not rank listings by it.