RightNow AI Releases AutoKernel Open-Source Framework for Autonomous GPU Kernel Optimization

RightNow AI has released AutoKernel, an open-source framework that uses an autonomous LLM agent loop to optimize GPU kernels for arbitrary PyTorch models. The framework targets GPU kernel optimization, described as one of the most grueling specializations in machine learning engineering, with the goal of automating this highly specialized task entirely. By applying autonomous agents to this problem, RightNow AI researchers aim to eliminate the need for manual GPU code writing. The open-source release makes this automation technology available to the broader ML engineering community.

AutoKernel Framework Release and Core Functionality

• ▪RightNow AI released AutoKernel as an open-source framework.
• ▪AutoKernel applies GPU kernel optimization to arbitrary PyTorch models.

Automation of GPU Kernel Optimization Using LLM Agents

• ▪Writing fast GPU code is one of the most grueling specializations in machine learning engineering.
• ▪RightNow AI researchers want to automate GPU code writing entirely.
• ▪AutoKernel uses an autonomous LLM agent loop for GPU kernel optimization.

Perspective of RightNow AI researchers

• ▪RightNow AI chose to release AutoKernel as open-source rather than keeping it proprietary.
• ▪RightNow AI researchers believe GPU kernel optimization can be fully automated using LLM agent loops.

1 source