Skip to content

Compute stack

Kernels and compilers

AI that writes GPU kernels, compiler passes and low-level code that train or serve AI faster.

ProjectGoogle DeepMind

Gemini models propose code changes
Evaluators keep best scoring versions
Finds better Gemini training kernels
Lowers cost to train next Gemini models
repeat

Kernels and compilersGoogle DeepMind · 2025

LLM agents write and benchmark kernels
Kernels speed up AI model training
Session data post-trains smaller kernel models
repeat

Kernels and compilersMeta · 2025

Mini-projectGitHub

TPU autoresearch

Coding agent edits LLM training code
Agent benchmarks code on TPU hardware
Verdict becomes context for next hypothesis
Wins raise training throughput for same models
repeat

Kernels and compilersAleksey Vlasenko · 2026

Mini-projectStanford CRFM

Stanford AI-generated kernels

Frontier LLMs write CUDA kernels
Fastest variants seed each new round
Output feeds next kernel-writing model
repeat

Kernels and compilersStanford CRFM (Ouyang, Mirhoseini, Liang) · 2025

ProjectCognition

QwQ-32B writes CUDA kernels
Model refines kernels using compiler feedback
RL rewards model on measured speedup
repeat

Kernels and compilersCognition, Stanford · 2025

DatasetHugging Face

KernelBook + KernelLLM

KernelLLM writes Triton GPU kernels
Faster kernels make AI training cheaper
Verified kernels train next kernel model
repeat

Kernels and compilersGPU MODE, Meta · 2025

Recently on the radar

PaperarXiv·28 Sep 2026·Kernels and compilers

AReaL-TIK: Stateful Agentic Optimization of Unified RL Kernels through an Optimization IR

The loop: The KernelBraid agent optimizes unified RL kernels through an intermediate representation and verifies the improvements. These faster kernels are then used to accelerate the reinforcement learning training process.

loop fit 9/10Ran Yan, Youhe Jiang, Jiayi Nie et al.via arXiv

Other loop types

Self-modifying agents Tools and harness Training data Self-reward and self-play Hardware and chips Inference efficiency Architecture and optimizer search Research automation AI R&D evals Interpretability and oversight