Rank pairs of AI-generated drug information responses as RLHF preference data, judging clinical accuracy, completeness, and appropriate sourcing against a scoring rubric.
We're seeking medical information/drug information specialists to produce preference-ranking data for RLHF by comparing AI-generated responses to HCP and patient drug queries, covering dosing, interactions, contraindications, and off-label use questions. You will judge which responses are most accurate, complete, and appropriately sourced, ranking them according to a defined rubric.
You will use authoritative drug information resources and clinical databases such as DrugBank, Micromedex, Lexicomp, FDA drug labeling, and PubMed to validate AI-generated responses and support evidence-based ranking decisions.
Your rankings will train the model's reward signal, so consistency and clear rationale matter as much as the final ranking.
A vetted network of doctors, scientists, and regulatory specialists who train and evaluate the models behind drug discovery, clinical trials, medical devices, and healthcare. Remote, flexible, matched to your field.