Publications

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

A comparative-analysis pipeline for using LLMs in bug bisection.

A profiling-guided framework for automated Triton kernel optimization.

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.

An early experiment exploring how large language models can assist static analysis.

A memory-management design for RISC-V enclaves.