The Modular Platform (includes MAX & Mojo)

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

Wave: Python Domain-Specific Language for High Performance Machine Learning

Nabla is a Distributed-Tensor and Scientific-Computing Framework; built from scratch on top of Mojo and MAX

Exercises for Learning MLIR (Originally written for PPoPP 2026)

Tensors and Dynamic neural networks in Python with strong GPU acceleration

A retargetable MLIR-based machine learning compiler and runtime toolkit.

Development repository for the Triton language and compiler

CUDA Tile IR is an MLIR-based intermediate representation and compiler infrastructure for CUDA kernel optimization, focusing on tile-based computation patterns and optimizations targeting NVIDIA te…

Learn GPU Programming in Mojo🔥 by Solving Puzzles🧩

The LLVM Project is a collection of modular and reusable compiler and toolchain technologies. Note: the repository does not accept github pull requests at this moment. Please submit your patches at…

Build an LLM from scratch with MAX

🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.

The best ChatGPT that $100 can buy.

Seemless interface of using PyTOrch distributed with Jupyter notebooks

A modern model graph visualizer and debugger

The Torch-MLIR project aims to provide first class support from the PyTorch ecosystem to the MLIR ecosystem.

A Python Compiler Design Toolkit

Distributed Triton for Parallel Systems

A minimal development of SSA theory

Embedded Universal DSL: a good DSL for us, by us

Run PyTorch LLMs locally on servers, desktop and mobile

Interactive deep learning book with multi-framework code, math, and discussions. Adopted at 500 universities from 70 countries including Stanford, MIT, Harvard, and Cambridge.

Efficient Triton Kernels for LLM Training

A collection of out-of-tree LLVM passes for teaching and learning