A US-based licensed physician contributes to AI training projects by evaluating and annotating AI-generated medical content, creating complex clinical scenarios, and assessing the accuracy and safety of AI responses. This is part-time, project-based work improving healthcare AI systems.
Pay not disclosed
The platform does not publish the pay for this job.
Open to residents of: United Arab Emirates, Argentina, Bulgaria, Bahrain, Brazil, Canada, Chile, Colombia, Czechia, Egypt, United Kingdom, Hungary, Indonesia, Israel, India, Jamaica, Jordan, Japan, South Korea, Kuwait, Lebanon, Mexico, Malaysia, New Zealand, Oman, Peru, Philippines, Poland, Qatar, Romania, Saudi Arabia, Singapore, Slovakia, Türkiye, United States, South Africa.
You apply on Mindrift, the AI training platform run by Toloka: submit your application for the role, pass a qualification step, then work on projects that match your expertise.
Official posting: https://apply.workable.com/j/0E6FB99F09
Last checked on October 8, 2026.
This remote, hourly contract asks you to review and grade AI-generated content about healthcare data, electronic health record systems, and clinical workflows. The work helps AI labs make their models more accurate on medical information. It suits health informatics professionals who can judge clinical and technical accuracy and explain their feedback in writing.
This remote contract role asks public health graduates to bring their domain knowledge to AI training. The work involves building realistic population health scenarios, writing evidence-based reference answers, and reviewing AI-generated health advice for accuracy, logic, bias, and ethics. It suits independent, self-directed specialists who can explain health data clearly, and no prior AI experience is required.
This role involves evaluating how well AI models reason through cardiology cases, identifying errors in clinical judgment that could harm patients. Cardiologists will write prompts to test AI reasoning, review AI outputs for accuracy against current guidelines, and provide correct answers when the model errs, serving as the training signal for AI improvement.
This role involves evaluating AI-generated responses to health and medical questions for clinical accuracy and safety. The expert identifies errors in reasoning, missed contraindications, unsafe advice, and writes correct answers grounded in current clinical practice, ensuring AI models provide guidance that real clinicians would trust.
This role asks practicing physicians to judge how well health AI models handle diagnoses, clinical reasoning and care plans. The work is part-time and asynchronous, aimed at MDs and DOs who currently see patients in primary care, internal medicine or hospital-based specialties.
Posted October 8, 2026
This short-term remote project asks practicing oncologists to test how frontier AI models reason through real-world cases, with a focus on clinical trial design, patient eligibility and treatment decisions. Contributors write challenging scenarios, judge model answers against current standard-of-care guidelines, and flag errors or unsafe advice. It suits clinicians with fellowship training and hands-on clinical trial experience who want to apply their expertise to AI research.
Posted October 8, 2026
