I’m currently a researcher working to advance our understanding of AI models. I enjoy building things, writing, and exploring this vast world. I’m broadly interested in decoding how large language models learn and research multimodal interpretability in the Darrell Group. I occasionally update a list of questions I’m exploring and thinking about.
Professionally, I was a early engineer at Brisk, where I helped build and scale an AI co-pilot for teachers to 100K users. I’ve also spent time working at Watershed on carbon data ingestion, Roblox on data infrastructure, and Thayermahan on autonomous maritime surveillance.
Previously, I studied CS and philosophy at UC Berkeley. Within the Berkeley ecosystem, I have:
- conducted NLP research to quantitatively analyze common challenges faced by product managers
- worked on auto-scaling and cloud support for Hydro, an open-source library to write and deploy distributed applications easily
- headed a client project with Recidiviz to improve sentencing outcome transparency for Codebase
- sourced sponsorships for Cal Hacks, the world’s largest collegiate hackathon
In my free time, I’m scootering and biking through cities, chilling in bookstores, and performing improv.