Physician (Resident/Fellow) AI Data Training

Help Train the Frontier of Medical AI — Residents & Fellows | $70/hr | Remote, Flexible

The most advanced AI models in the world are now being used by millions of people for health questions every week. They are astonishingly capable — and they still get clinical reasoning wrong in ways that only a trained clinician can catch. We're recruiting US residents and fellows to be part of the team that finds those failures and fixes them.

This is not data labelling or chart review. You'll be working directly with the outputs of frontier large language models — the same class of system behind the AI tools reshaping medicine — and your job is to think harder than the model did.

What the work actually involves:

• Reading a model's response to a clinical scenario and rigorously assessing it against a detailed, criterion-by-criterion rubric: did it reach the right diagnosis, did it order the right investigations, did it miss a red flag, did it state the safety-critical step explicitly or only imply it?
• Identifying where and why the reasoning breaks down — distinguishing a model that got the right answer for the wrong reason from one that reasoned soundly but omitted a must-state safety action.
• Writing clear, structured justifications for each judgement, so your clinical reasoning becomes a training signal the model learns from.
• Comparing multiple models' answers to the same case and articulating what separates a strong clinical response from a dangerous one.

Why it matters: every correction you make becomes part of how these systems reason for millions of future users. When you catch a model confusing a diagnosis with its cause, or skipping the bedside step that confirms it, you're preventing that error at scale.

You are: a current US resident or fellow in any specialty, comfortable working through clinical cases and — crucially — able to explain why an answer is right or wrong, not just that it is. Strong reasoning and attention to detail matter more than seniority at this tier.

The work: fully remote, genuinely flexible. Work between rotations, in the evenings, or at weekends — no fixed shifts, no on-call, no minimum commitment. Most contributors do 5–15 hours a week.

Rate: $70/hour- paid weekly

To apply: send your CV and you will be contacted within 24 hours if shortlisted.