Mindrift ir mākslīgā intelekta talantu tirgus un SaaS platforma, kas savieno profesionālus darba veicējus un konkrētu jomu ekspertus ar tehnoloģiju uzņēmumu mākslīgā intelekta apmācības un novērtēšanas projektiem. Kā Toloka daļa tā sadarbojas ar rakstniekiem, redaktoriem, kvalitātes nodrošināšanas speciālistiem, zinātniekiem un citiem jomu ekspertiem, kuri palīdz veidot apmācības datus, novērtēt modeļu drošību un spriestspēju, izvērtēt multivides kvalitāti un radīt reālās pasaules scenārijus mākslīgā intelekta veiktspējas un uzticamības uzlabošanai. Mindrift piesaista un pārbauda darba veicējus, izmantojot novērtējumus un identitātes verifikāciju, pēc tam piedāvājot elastīgas, apmaksātas projektu iespējas mākslīgā intelekta apmācības un novērtēšanas jomā.
Remote Clinical AI Evaluator (MD/DO)
US-based physicians review electronic health record cases and clinical reasoning to help assess AI performance. The project includes diagnosis and treatment benchmarks, with an estimated 10–20 hours of work per week during active phases.
Apraksts
- Review electronic health record vignettes, questions, proposed answers, and distractor answers for diagnosis and treatment tasks.
- Assess cases across cardiovascular, nervous, hematologic, respiratory, digestive, urinary, reproductive, musculoskeletal, and integumentary systems.
- Rate clinical reasoning, case accuracy and completeness, answer correctness and gradeability, and distractor quality.
- Check whether each conclusion follows from the facts provided in the vignette.
- Revise reasoning traces and provide concise explanations.
- Complete three independent blind reviews, followed by physician adjudication.
- Work on project tasks during active phases, estimated at 10–20 hours per week.
Prasības
- Be a US-based, actively practicing physician with an MD or DO.
- Hold an active, unrestricted US medical license, verified with the issuing medical board.
- Have board certification or completed residency training.
- Have recent direct patient care experience; attending-level experience is preferred.
- Bring broad diagnostic and treatment experience across multiple systems in internal, family, emergency, or hospital medicine.
- Demonstrate strong clinical reasoning in differential diagnosis, next-step management, and applying evidence-based guidelines.
- Experience in medical AI evaluation, clinical content or exam-question review, or medical annotation or quality assurance is a plus.
- Have strong written English proficiency (C1 or higher).
- This role is not suitable for medical students, unlicensed medical graduates, non-physician clinicians, or non-clinical healthcare professionals.
- Submit a CV in English and indicate English proficiency.
- Meet task deadlines and the stated acceptance criteria.
Priekšrocības
- Participation is project-based.
- Projects are part-time and non-permanent.
- Payment is per accepted task, up to the equivalent of $130 per hour depending on qualification tier and task efficiency.