Published multiple papers including 'Evaluating scientific theories as predictive models in language neuroscience', 'Mixture of Inputs', 'Interpretable Language Modeling via Induction-head Ngram Models', and more; involved in projects such as OmniGuard and systematic bias assessment in clinical decision instrument development.
Research Experience
Senior researcher at Microsoft Research (Deep Learning Group); extensive research experience in the interpretability of neural networks, fMRI responses to language, and clinical decision rules.
Education
PhD from UC Berkeley, advised by Prof. Bin Yu.
Background
Research interests include interpretability methods, semantic brain mapping, and clinical decision rules. Currently focused on LLM interpretability, augmented iModels, attention steering, explanation penalization, and adaptive wavelet distillation.
Miscellany
Personal interests include open-source ML, causal inference, complexity theory, disentanglement, deep learning theory, and more.