Derek Fitzgerald

Key Skills

Software

Mercor

Top Subject Matter

No subject matter listed

Top Data Types

Document

Medical Dicom

Text

Top Label Types

Prompt Response Writing SFT

Freelancer Overview

I am a board certified PM and R physician with fellowship training in interventional pain medicine, and I bring both clinical and research level experience in working with complex medical data. Throughout my training and practice, I have designed and coordinated research projects that required detailed data collection, labeling, and interpretation, including studying racial disparities in pediatric treatment outcomes and evaluating immersive virtual reality for procedural pain control. This work required precision, structured thinking, and a strong understanding of medical terminology and clinical context. In addition, I have hands on experience training AI models through my work with Mercor, where I evaluated, scored, and annotated clinical and medical outputs for accuracy, reasoning quality, and real world applicability. This experience strengthened my ability to apply strict evaluation criteria while maintaining contextual awareness and domain expertise. My background as a practicing physician combined with direct AI training work makes me well equipped to contribute meaningfully to healthcare focused data labeling and AI development projects.

Labeling Experience

AI Training Prompt Writer

MercorMedical DicomPrompt Response Writing SFT

During my time with Mercor, I focused on creating adversarial medical prompts designed to induce model failure in clinical reasoning and safety. I developed realistic, high complexity scenarios that tested diagnostic accuracy, treatment decisions, contraindications, and risk assessment, specifically aiming to expose hallucinations, unsafe recommendations, and flawed reasoning. The project operated at scale with structured review processes, including calibration and secondary validation to ensure prompts were clinically accurate, sufficiently challenging, and effective at differentiating model performance. This work required deep medical expertise, precision in language, and a clear understanding of how AI systems fail in healthcare contexts.

During my time with Mercor, I focused on creating adversarial medical prompts designed to induce model failure in clinical reasoning and safety. I developed realistic, high complexity scenarios that tested diagnostic accuracy, treatment decisions, contraindications, and risk assessment, specifically aiming to expose hallucinations, unsafe recommendations, and flawed reasoning. The project operated at scale with structured review processes, including calibration and secondary validation to ensure prompts were clinically accurate, sufficiently challenging, and effective at differentiating model performance. This work required deep medical expertise, precision in language, and a clear understanding of how AI systems fail in healthcare contexts.

2025 - 2025

Education

Work History