RightNow AI Releases AutoKernel: An Open-Source Framework that Applies an Autonomous Agent Loop to GPU Kernel Optimization for Arbitrary PyTorch Models

RightNow AI released AutoKernel, an open-source framework that uses an autonomous LLM agent loop to optimize GPU kernels for PyTorch models.

The agent edits Triton kernel files, runs correctness and performance benchmarks, and keeps or reverts changes based on the results.

AutoKernel uses git-backed experiment tracking and a long instruction file to guide optimization behavior.

The goal is to reduce manual kernel tuning work and make GPU performance improvements more accessible.