Walter Gerych

Walter Gerych is a tenure-track Assistant Professor of Computer Science at Worcester Polytechnic Institute (WPI). He earned his PhD in Data Science from WPI, advised by Elke Rundensteiner and Emmanuel Agu, and subsequently was a postdoctoral associate in the Healthy ML lab at MIT CSAIL under Marzyeh Ghassemi before returning to WPI as faculty.

His research centers on trustworthy machine learning: removing harmful behaviors and spurious correlations from models, improving fairness, advancing AI safety, and developing robust measures of confidence and uncertainty that indicate when a model's prediction can be trusted. A recurring theme of his recent work is debiasing large pre-trained and multi-modal foundation models using efficient, minimally invasive approaches that preserve downstream performance, with frequent application to machine learning for healthcare and generative AI. At MIT he led an SERC group on bias in clinical large language models. His work has appeared at top venues including NeurIPS, ICLR, AAAI, ACL, and CIKM.

Gerych is also active in service to the research community. He has served as an organizer for the Time Series for Health workshop, as a Senior Area Chair for the Models and Methods track at the Conference on Health, Inference, and Learning (CHIL 2024), and as an Area Chair for NeurIPS (2025).

At the camp: PM Lecturer — Day 4 Methods & Evaluation II, 11:00 AM Lecture (Methods & evaluation paper spotlight).