RLinf: Reinforcement Learning Infrastructure for Embodied and Agentic AI

Model interpretability and understanding for PyTorch

The WeightWatcher tool for predicting the accuracy of Deep Neural Networks

A flexible and high-performance training framework designed for large-scale foundation model training on AMD GPUs

High-performance GPU kernels for LLM inference in OpenAI Triton. Fused RMSNorm, SwiGLU, INT8 GEMM with benchmarks and roofline analysis.

Code release for book "Efficient Training in PyTorch"

AI 基础知识 - GPU 架构、CUDA 编程、大模型基础及AI Agent 相关知识。

TradingAgents: Multi-Agents LLM Financial Trading Framework

AI Infrastructure Engineer Learning Track - Production ML infrastructure curriculum (2-4 years experience)

Academic Research Skills for Claude Code: research → write → review → revise → finalize

Now, Stronger AI Pushes Frontiers, Stronger Our Shared Future.

ARIS ⚔️ (Auto-Research-In-Sleep) — Lightweight Markdown-only skills for autonomous ML research: cross-model review loops, idea discovery, and experiment automation. No framework, no lock-in — works…

A curriculum for learning about gpu performance engineering, from scratch to what the frontier AI labs do

AI Infrastructure Performance Engineer Learning Track - GPU optimization, inference optimization, and cost reduction

Universal LLM Deployment Engine with ML Compilation

An Extensible Deep Learning Library

Helpful kernel tutorials, examples and SKILLs for tile-based GPU programming

Kernels, of the mega variety :)

Tile primitives for speedy kernels

LMCache: Supercharge Your LLM with the Fastest KV Cache Layer

CUDA Core Compute Libraries

PyTorch compiler that accelerates training and inference. Get built-in optimizations for performance, memory, parallelism, and easily write your own.

A systematic and pedagogical way to derive the correctness structure of 2D Register Allocated GEMM before coding.

DeepSeek 4 Flash and PRO local inference engine for Metal, CUDA and ROCm

Reverse engineering NVIDIA SASS instruction dictionary, kernel audits and pattern recognition across GPU architectures.

Fast Polar Decomposition for Muon

Skills for Real Engineers. Straight from my .claude directory.