Autoresearch for GPU kernels. Give it any PyTorch model, go to sleep, wake up to optimized Triton kernels.

Production-Grade Autoresearch. Ideal for GPU kernels, ML model development, feature engineering, prompt engineering, and other optimizable code.

Custom Linux kernels purpose-built for Apple Mac hardware

Forge: Swarm Agents That Turn Slow PyTorch Into Fast CUDA/Triton Kernels

Open source skill library for AI coding agents to write, optimize, and debug high performance compute kernels across CUDA, Triton, and quantized workloads.

Learn Triton by building FlashAttention from scratch — V2 kernels, persistent threads, mask DSL, profiling toolkit, bilingual docs

Automatic Triton kernel generation and optimization for Intel GPU, powered by Claude Code.

Extended TileLang as a unified DSL to enable high-performance kernel development for Near-Memory Computing, Distributed Memory AI Accelerators, and Networked Accelerators.

Noeris — autonomous kernel fusion discovery + Triton autotuning for LLM kernels and Gemma layer deeper fusion (A100/H100 wins).

Enable autonomous AI agents to optimize LLM training code through iterative experiments and improve models without manual intervention overnight

A collection of high-performance CUDA kernels and experiments for learning and optimizing GPU compute primitives.

CUDA kernel optimization lab: GEMM, FlashAttention, quantization, and GPU performance learning.

The Ultimate Kernel Orchestration Suite for Windows. Optimized for low-latency development and high-priority workloads.

Optimized Ubuntu Touch for Lenovo Tab M8 HD (TB-8505F) - Kernel improvements, performance tuning, boot experience, and system optimizations for the MediaTek Helio A22 tablet

Hand-tuned NVIDIA SASS kernels for RTX 3070 Ti (GA104, sm_86): 31,910 GFLOPS HGEMM, 41,721 dense-equiv 2:4 sparse, 11,453 GFLOPS Flash Attention, no cuBLAS / cuDNN / PyTorch. 6-chapter tutorial + Chladni-pattern memory layout study.

Optimize PyTorch GPU kernels by autonomously profiling, extracting, and improving Triton or CUDA C++ code for better performance and efficiency.

Optimisation HPC et analyse de performances d'un noyau de calcul pour le compilateur GCC avec MAQAO.

10,000-image LeNet-5 forward pass in ~28 ms on a single A40 via fused convolution and Tensor Cores (TF32).

Systematic CUDA kernel engineering from SGEMM fundamentals to reusable kernels, advanced optimization, and inference components

Benchmark plots and performance analysis for upstream pto-kernels development