Using LLMs to Optimize CI Performance and Analysis

This talk explores how to leverage Large Language Models (LLMs) to enhance Continuous Integration (CI) workflows and regression analysis. Learn how to quickly implement off-the-shelf LLMs like Anthropic's Claude Sonnet 3.5 in software development processes, with practical examples using LLVM buildbots. Discover techniques for efficiently identifying problematic commits without time-consuming bisection processes, analyzing commits when investigating unusual regressions, and understanding both current limitations and future possibilities for LLM integration in performance analysis and CI automation for open source projects.