What the work actually is

Frontier labs cannot generate clinical judgment internally, so they buy it. On a typical project you might author a case drawn from your own practice — the presentation, the workup, the constraints a real clinic imposes — and then write the reference answer a model's attempt will be graded against. Other projects are pure evaluation: read a model's differential and say whether it is ordered sensibly, whether the guideline it cites actually applies to this patient's age, renal function and comorbidity burden, and whether the plan is safe for someone to carry out. The highest-value task is usually the third one — flagging the recommendation that is defensible in a textbook and wrong in front of the patient, given access, insurance, or what the care setting can deliver.

What the screen is looking for

  • A verifiable clinical credential and real practice experience. Credentials are checked; this is not an honour-system pool.
  • Written reasoning. Almost every deliverable is prose explaining why an answer is wrong, not a score alone. Graders who write "incorrect" without the mechanism produce unusable data.
  • Tolerance for ambiguous instructions, and the habit of flagging them rather than guessing. Rubrics arrive half-formed and get sharpened by the people using them.

The pool is deliberately broader than physicians: internal medicine and surgical specialties, diagnostics, psychiatry and clinical psychology, nursing, dentistry, pharmacy, laboratory and medical technology, public health and epidemiology, healthcare administration, and medical writing.

Logistics, stated plainly

Remote and asynchronous. You apply once with a resume and your work location, take an AI interview of roughly twenty minutes, and — once verified — sit in the medicine pool. This listing issues no accept or reject decision. Invitations arrive when a client project needs your profession or specialty, and that can be a week out or several months. Rate, hours, and client are named on the project listing, and any hiring decision happens there. Most project work is part-time and fits around clinical hours; the observed $60–180 band reflects posted rates across the field and is not a guarantee for any individual engagement.