About me
I am a Ph.D. candidate in the Department of Statistics at UC Berkeley, advised by Ryan Tibshirani and Giles Hooker. Before coming to Berkeley in 2021, I completed my B.S. at Yale University, where I did NLP research with the late Dragomir Radev.
In summer 2024, I interned in the computer vision group at Apple, designing interpretability methods to analyze failure modes of FaceID. The year prior, I worked on Digital Pathology at Genentech, hosted by Patrick Kimes. I also spent a summer in undergrad at a deep learning startup in Israel.
Research Interests
My research aims to ensure the safe and effective deployment of machine learning. To that end, my work on black-box interpretability enhances the reliability of SHAP and LIME, and introduces new techniques to understand image models. I am also interested in uncertainty quantification, especially for rankings and time series.
While I am broadly interested in ML for health, my current applied emphasis is in epidemiology. Through the Delphi Group, I construct analytical and algorithmic tools to track infectious disease risks as they unfold in real-time.
Please feel free to reach out if you’d like to connect!
News
- June 2025. Unifying Image Counterfactuals and Feature Attributions with Latent-Space Adversarial Attacks has been accepted for the Actionable Interpretability Workshop at ICML.
- May 2025. Statistical Significance of Feature Importance Rankings has been accepted for publication at UAI. Our methods stabilize SHAP, LIME, and other interpretability methods using ideas from Gaussian Rank Verification.
- Jan 2025. Our new stats preprint, Gaussian Rank Verification, is on Arxiv.
- Dec 2024. Challenges in Estimating Time-Varying Epidemic Severity Rates from Aggregate Data is up on MedRxiv.
- Dec 2024: I passed my Qualifying Exam! Thanks to my committee: Ryan Giordano, Will Fithian, Chris Paciorek, Ryan Tibshirani, and Giles Hooker.