Back to portfolio

AI Kernel Log Analysis

Proposed and led development of an AI-driven tool to assist AMDGPU kernel developers in triaging errors from Linux kernel logs — reducing manual effort and improving issue resolution speed across IQE, CVS, and KMD teams.

Key Contributions

Leadership Buy-In

Proposed the AI log analysis concept and secured leadership approval for cross-team integration with IQE, CVS, and Linux kernel development teams.

Go Parsing Framework

Designed and implemented a Go-based framework to parse and categorize AMDGPU kernel logs before passing structured data to an LLM for actionable insights.

LLM Integration Workflow

Built end-to-end workflow: kernel log capture → structured parsing → LLM analysis → developer-facing triage recommendations.

Developer Productivity

Reduced manual log triage effort for kernel developers during system bring-up and post-silicon validation of MI-Series GPUs.