Read skin. Teach the AI. Get paid for your expertise.
Autoderm is recruiting doctors, residents and medical students worldwide as Clinical Reviewers. Remote, paid, and built around the schedule you already have.
Apply now
You read real skin images and record what you see. Your judgement becomes the reference standard our AI learns from, and the weekly feedback makes you a sharper diagnostician while you do it.
What you decide
You review real smartphone images of skin disease and record two things: the most likely diagnosis, and what the patient should do next: urgent care, routine review, self-limiting, or self-care.
That is the whole task. No patient contact. No consultations. No letters, no reports, no on-call. Nothing to dictate and nobody running late. You work through cases at your own pace, from anywhere, whenever it suits you. Six seconds if you are sure. Longer if the case is interesting.
The images come from more than 160 countries, many with no dermatologist to wait for. What you read here shapes what the model can recognise for people your clinic will never see.
Please note: this is not a teledermatology consulting role. You are not diagnosing or advising patients. If online patient consultations are what you are looking for, say so in the form and we will point you to the right programme.
It is sudoku for doctors who like skin
Ask a dermatologist why they chose the specialty and you get some version of the same answer: the pattern recognition is satisfying. A rash resolves into a diagnosis. Something clicks.
And it is not one-way
Most doctors practise for decades without ever learning how often they were right. A clinic has no feedback loop. The patient improves, or does not come back, and the answer never returns to you.
Here it returns every week. A dashboard shows where you and the two other clinicians who read the same cases agreed, where you did not, and what the consensus was, with notes on the features you may have weighted differently. Your agreement rate is tracked by condition, body site, skin type, and over time.
That is the part that changes your reading. Hundreds of cases a month, on presentations you might otherwise see twice a year, with the answer coming back each time.
This is not just a nice idea. It is measurable
Repeated image diagnosis with feedback is one of the better-evidenced techniques in medical education.
Student diagnostic accuracy
Controlled study, repeated dermatology image modules
Decision time
Same study. Faster and more accurate at once
Against classroom teaching
Randomised dermoscopy self-learning research
If you are a student, resident, or a doctor outside dermatology, expect to get better.
Every image is read three times
You are never the only person looking at a case. Each image is read independently by three healthcare professionals, and the reference label comes from their agreement, not from any single reader. No reading is overruled by a job title.
Reader 1
Reader 2
Reader 3
So you can apply without feeling you must be certain on every case. Uncertainty is information, and there is a button for it. Flagging a case as unclear is a useful answer, not a failure.
Higher-stakes cases, including anything suspicious for skin cancer, always include qualified clinicians among the readers, and any positive flag is escalated for specialist review.
Who we are looking for
We select on demonstrated diagnostic accuracy, not on job title.
A dermatologist working mainly in cosmetics may read medical cases less accurately than a family physician who sees skin complaints every day. So we do not screen on credentials. We screen on how you perform on real cases.
A dermatologist, at any career stage
A dermatology resident or specialty trainee
Any qualified doctor with dermatology training, a skin-heavy caseload, or a serious clinical interest. General practice, family medicine, internal medicine, paediatrics and emergency medicine reviewers all read for us
A medical student in your clinical years with a genuine interest in dermatology
You do not need a formal dermatology qualification. You do need working English (our diagnostic taxonomy is in English) and a reliable internet connection. Dermoscopy experience is welcome but not required.
Your CV and the screening test decide, not your title.
If you are early in your career
This is unusual experience: thousands of real cases from more than 160 countries, across every skin type, measured against expert consensus. Few training posts offer that volume. Almost none offer the feedback loop.
Countries represented
Across every skin type
Of participation
Documenting your hours and case volume
In dataset publications
Where you have contributed substantially
It is paid work, not an unpaid internship.
How it works
We select on demonstrated diagnostic accuracy, not on job title.
A dermatologist working mainly in cosmetics may read medical cases less accurately than a family physician who sees skin complaints every day. So we do not screen on credentials. We screen on how you perform on real cases.
01
Apply
Complete the form below and paste in your CV. Around three minutes.
02
Take the screening test
Twenty cases, about thirty minutes, in one sitting. Short answers are fine. It is genuinely difficult, and we do not expect perfect scores. We benchmark against published agreement rates between specialists, not against an impossible standard. Some reviewers without formal dermatology training outperform specialists on it, which is the entire reason we test rather than assume.
03
Onboarding
If you qualify, we agree your weekly hours and your rate, and walk you through the platform. Your first cases are read alongside an experienced reviewer.
04
Start reading, start earning
A monthly retainer against committed weekly hours, not per-image piecework. Choose 5, 10 or 20 hours per week. You know what you are earning before you start.
Who we are
First Derm has provided online skin consultations since 2014, answering cases from patients worldwide within hours.
Autoderm is the AI built on that foundation: an API that healthcare platforms integrate to return informational skin condition suggestions, developed and maintained as a regulated medical device. It is CE-marked (MDD Class I legacy, MDR Class IIa in progress), holds FDA Breakthrough Device Designation, and its primary care validation is published in Nature Scientific Reports.
Top-1 accuracy
Boots UK GP Reader Study, 2024
Dermatology referral reduction
Boots UK GP Reader Study, 2024
The model exists to carry specialist pattern recognition into the places that have none.
It is designed to support clinicians, never to replace them, and that is structural rather than reassuring. A model built to defer to clinical judgement has to be built on clinical judgement, and measured against it. That is the work.
Apply to become a Clinical Reviewer
We reply to every application. If you look like a good fit, you will have the screening test within a few days.