Auto GPU Kernel
github.com/dogacel/auto-gpu-kernelAbout
Auto GPU Kernel is an autonomous agent system that discovers and optimizes GPU kernels, evolving both the kernel code and its benchmarking harness rather than relying on a human-authored one. It won the agent-only track of the MLSys 2026 FlashInfer AI Kernel Generation Contest for DeepSeek Sparse Attention, achieving an average 34.93x speedup, with its kopt tool driving optimization and kbench handling validation and benchmarking.
Why it made the leaderboard
Gives engineers a validated meta-harness for automating GPU kernel optimization, an area where agentic code generation has produced a documented, contest-winning speedup rather than an unverified demo claim.
Tags
gpukernel-optimizationai-agentsmlsysflashinferautonomous-agent
Tech Stack
Python
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.