Building Trustworthy GenAI for Clinical Data | Hands-on Session

Join us for a special online hands on session with Maya Malamud - an AI Consultant & Data Scientist, helping healthcare organizations build trustworthy AI systems.

This is a collaboration between the MeDS community and Women in Data-Science (WiDS) community!

What to expect?
Most GenAI systems impress in demos but fail quietly in real clinical settings. In healthcare, this isn’t just inconvenient — it risks patient safety and erodes trust.
In this hands-on session, we’ll build a minimal GenAI pipeline tailored to a medical data context and learn how to evaluate, monitor, and improve its reliability using open-source tools such as LangFuse and RAGAS.
Participants will see how seemingly simple clinical queries can trigger hallucinations, inconsistencies, and retrieval failures — and how to detect these issues before they reach clinicians. We’ll cover practical techniques for measuring hallucinations, identifying performance drift as clinical data evolves, and setting up basic improvement loops that move a prototype toward something clinicians can safely rely on.
You’ll walk away with reusable code, evaluation frameworks, and a deeper understanding of how to build GenAI tools that meet the reliability expectations of medical environments.

Looking forward to see you!

Sponsored by Avon AI