Projects

SecureLLM: Using Compositionality to Build Provably Secure Language Models

Research paper on building provably secure language models for private, sensitive, and secret data. Third author on this work that introduces a new approach to LLM security through compositional fine-tuning.

Research & Technical WritingAI SafetyLLMs

CS350: Exploration of AI Safety

A culminating report on my exploration of AI safety considerations over my independent study the fall of my senior year.

Research & Technical WritingTech to Policy

Personal Website

You're here! More coming soon.
Built with Next.js and deployed on Vercel.

Web DevelopmentGit