Hi! My name is Divya. I am a National Library of Medicine Fellow at Columbia’s Department of Biomedical Informatics.

I develop machine learning methods that use large-scale healthcare datasets to understand how people seek, receive, and deliver care, and to develop more reliable AI systems. My recent work develops methods to audit training data (NWH 2024; JAMA HF 2026); to evaluate models when labels are limited or unreliable (CHI 2023; NeurIPS 2025); and to improve the reliability of pretrained systems at test time (ICCV 2021; CVPR 2025). My current research explores how AI can support care throughout pregnancy.

I am always excited to work with students and collaborators on problems where clinical settings raise new technical questions. Previously, I was a postdoc at Cornell Tech, advised by Emma Pierson and Jenna Wiens. Before that, I earned my Ph.D. at MIT, where I was lucky to work with John Guttag. You can reach me at dms2351@cumc.columbia.edu.