Medical Domain Expert — LLM Training & Evaluation — Atlas Capture
As a Medical Domain Expert, I evaluated 100–500 medical LLM outputs using structured rubrics focused on clinical accuracy, safety, dosing, and citation fidelity. I provided expert clinical judgment, produced written rationales for rubric-based scores, and generated reward signals for reinforcement learning from human feedback (RLHF). I fact-checked AI outputs, flagged unsafe recommendations, and contributed by authoring prompts spanning pharmacology and clinical decision-making. • Evaluated large volumes of AI-generated medical text for accuracy and safety • Applied licensed-pharmacist expertise to refine LLM outputs • Used structured rubrics to standardize medical content assessment • Authored and edited medical prompts for advanced model probing