Live, remote AI-training roles relevant to Health, updated daily. Applied Clinical Judgement is a UK-based referral intermediary: we point you to genuine openings on the major training platforms and are paid only when a referral succeeds. Pay rates are shown on each role; we never display our referral fee.
6 live Health roles · updated daily
Certified Health & Wellness Coach Professionals
Turing seeks certified health and wellness coaches to evaluate AI-generated coaching conversations for accuracy, safety, and empathy. You'll assess large language model outputs against evidence-based coaching frameworks, annotate quality dimensions, and flag unsafe or misleading advice. The role suits professionals with coaching credentials and understanding of behaviour change principles. This is an 8-week freelance assignment with flexible remote working.
Sleep Specialist / Behavioral Sleep Medicine Professional
Turing seeks sleep specialists to annotate and evaluate sleep health data for AI model development. You'll classify sleep stages, duration, and disturbances from sensor data and logs, flag anomalies, and apply standardized annotation guidelines. Background in sleep science, physiology, or related health field required, plus familiarity with sleep tracking technologies. Eight-week contractor assignment, fully remote.
Pharmacist
$70.0-$70.0 / hourly. An AI research lab is recruiting licensed US pharmacists to evaluate medication-related clinical questions and review AI-generated prior authorisation responses for accuracy, safety, and completeness. The role involves applying clinical expertise to validate AI systems, correct inaccurate outputs, and flag patient safety concerns. Work takes place on the Mercor platform across 30–40 hours weekly.
Healthcare Expert
$90.0-$100.0 per hour. Mercor seeks licensed MDs with 2+ years clinical experience to evaluate AI-generated healthcare content for accuracy and quality. Work involves reviewing written material, answering domain-specific questions, and providing feedback to improve AI reasoning in medical contexts. Generalist clinical background preferred over narrow subspecialty. Teaching or training experience beneficial. Approximately 10 hours weekly, fully remote, US-based applicants only.
Medical Specialist
$20–$40 / hr. micro1 seeks medical practitioners, researchers and documentation specialists to contribute original, expert-authored materials for AI training. You'll submit existing professional documents—case reports, clinical records, research manuscripts—rich in text, charts and diagrams. Work involves applying professional judgment to de-identify sensitive information. Advanced medical credentials and extensive documentation experience preferred. Remote contractor role.
Health Care Specialist
$15-$25/hr. Health Care Specialist role on micro1 for those with healthcare, clinical support, or patient care expertise. You'll review medical documentation, analyse clinical workflows, and ensure compliance with healthcare standards. No AI experience required—your domain knowledge drives the work. Contractor position, remote, part-time.
No live roles match your search.
AI training work is organised by profession, task and software — not by topic or sector. Try your field (for example “nursing” or “Python”), clear the filters, or browse the categories further down the page. The always-open talent pools below are a good place to start.
“More than 40 million people turn to ChatGPT every day with healthcare questions. One in four of our 800 million regular users submits a health-related prompt every week.”
— OpenAI, AI as a Healthcare Ally, January 2026 · Read the full report
Health AI can’t be trained safely on data alone
Health AI systems are increasingly used to support clinical decision-making, triage, documentation, and patient communication. Training them responsibly means involving clinicians directly — particularly where nuance, uncertainty, and patient safety are at stake.
Without clinical judgement embedded in training, AI systems can appear confident while being unsafe, miss real-world trade-offs, and optimise for plausible-sounding answers rather than clinically appropriate ones. That’s why health AI projects now rely on clinicians reviewing prompts, assessing outputs, and defining what “good” looks like in practice — carried out remotely, task by task, through platforms like Mercor and Micro1.
You won’t be coding or building AI systems. Tasks involve reviewing AI-generated responses to clinical questions, scoring diagnostic reasoning, flagging unsafe or misleading outputs, and defining reference answers for specific clinical scenarios. Your clinical reasoning, risk awareness, and contextual judgement are the qualification — no technical background needed.
Explore how remote AI training work is done
What the work actually involves — from a platform’s own medical team
The clearest inside view of this work we’ve found comes from micro1 itself. In the session below, their director of medical research — a doctor who’s also a computer engineer — and a Harvard-trained MD walk through how clinicians train and evaluate health AI, including a real pathology case study. It’s about half an hour. Worth it before you apply.
Two failure modes only clinicians catch
Health AI fails in ways a generic reviewer wouldn’t notice. micro1’s team describe two that come up constantly.
Clinical overreach. Their example: a model reviewing a stomach-biopsy pathology report correctly identified H. pylori — then volunteered a treatment plan. It was only asked for the diagnosis. Therapy decisions depend on the whole patient, so an unprompted medication recommendation isn’t helpfulness, it’s a safety failure. Spotting that distinction is clinical judgement, and it’s precisely what these roles pay for.
Unsupported inference. Models tend to fill gaps in clinical data with plausible assumptions rather than saying the evidence isn’t sufficient. In most domains, filling gaps is useful. In medicine, hedging — asking for more examination, more history, more results — is the correct behaviour. Models have to be taught it by people who manage uncertainty for a living.
The tasks you’d actually be doing
The session sets out the day-to-day task types in plain terms. Grading single AI responses against structured rubrics. Reviewing multi-turn conversations — does the model hold context the way you would while taking a history? Annotating images, such as drawing bounding boxes around fractures on X-rays. Grading model interpretations of lab reports and pathology documents. Writing “golden responses” — the answer an expert would give, used as the training target. And reviewing the model’s reasoning path, not just its final answer: an output can be right for the wrong clinical reasons, and that still needs correcting.
Quality control looks familiar too. In one project they describe — building a detection system for drug-induced liver injury from pathology slides — every slide was independently reviewed by two experts, with disagreement between reviewers treated as a signal in itself.
Why the demand isn’t going away
Models now handle textbook presentations impressively — one of micro1’s physicians calls them “savant clinicians”. What they can’t yet do is real clinical workflow: knowing what to ask next, working with incomplete information, staying current. With one to three million life-science papers published every year, and models drifting as they’re retrained, evaluation is continuous rather than one-off. micro1 says its network includes over 200,000 verified physicians across essentially every specialty, with onboarding onto a project typically inside 48 hours — their figures, not ours, but a fair indication of the scale of demand.
The live health roles above are where that demand shows up.
Who this work suits
This work suits clinicians with real-world healthcare experience. Doctors — especially GPs, family doctors, and psychiatrists — are in high demand, as are specialists including radiologists, cardiologists, and neurologists. Psychologists, nurses, advanced practitioners, pharmacists, physician associates, and allied health professionals such as physiotherapists and occupational therapists are all eligible. Work is flexible and task-based — compatible with portfolio or locum careers, subject to your contracts and employer policies.
Find out if this suits your background
Mercor is the highest-paying platform we refer to — the two evaluator roles above sit at $80–$120/hr, the US physician role at $130–$180/hr. Here’s how their apply-once model works, from the co-founder who built it.
Common questions
Is this work remote? Yes. Health AI training and evaluation work is carried out remotely, in your own time, at times of your choosing.
Do I need technical or coding skills? No. Clinical judgement and real-world experience are what you need. Most tasks involve reading, reasoning, and writing — assessing AI outputs, defining reference answers, and quality-checking responses.
Is patient data involved? No identifiable patient data is used.
Is this employment or freelance contract work? Contract project work, not employment. Most arrangements are flexible and task-based rather than salaried. You’ll be contracting directly with Mercor or Micro1, not with this site.
Can I do this alongside NHS or locum work? In many cases yes — subject to your employer’s policy on secondary work and any conflicts of interest.