Principled LLMs for reliable systems software.

I develop LLM-assisted program analysis techniques to analyze and secure complex systems software. A central line of my Ph.D. research is The Hitchhiker's Guide to Program Analysis, a three-part research series on practical bug detection for the Linux kernel and related systems, with the first two parts published at OOPSLA and ASE. Across this series, I explore how to leverage LLMs' understanding of code and broader software knowledge to make program analysis more precise and reliable.

More broadly, my goal is to make LLM-assisted software engineering genuinely principled. Better models alone cannot close the gaps between what developers intend, what LLMs assume, and what programs actually do. I study how to bridge these gaps—combining program analysis and LLMs to understand, verify, and ultimately reshape complex software so that strong reliability guarantees become practical at scale.

We discuss a more principled approach to integrating LLMs (i.e., mostly harmless LLMs) with static analysis for practical bug analysis in large codebases.

BugLens guides LLMs through structured reasoning steps to post-refine taint-style static analysis reports in the Linux kernel.

LLift integrates LLMs with static analysis to improve practical bug detection in large codebases such as the Linux kernel.