I develop reliable language models and natural language processing methods for medicine. At HMS/BCH, I study clinical prediction from electronic health records, with an emphasis on confidence estimation, explanation generation, and reasoning over long clinical documents.
My work spans the language-model lifecycle, from biomedical pre-training to clinical adaptation and evaluation. I received my Ph.D. in Computer Science from Korea University, where I co-led BioBERT, and collaborated with AstraZeneca from 2020 to 2023 on bringing biomedical NLP into industrial use.
Before NLP, I spent much of my undergraduate years studying information security and organizing student-led InfoSec events. I learned a lot from my peers and enjoyed sharing what I knew with junior students. What I learned during those years continues to help me when working with clinical data, where privacy and security are essential—not something I ever expected back then!