Official Implementation of EAGLE-1 (ICML'24), EAGLE-2 (EMNLP'24), and EAGLE-3 (NeurIPS'25).

Notes and exploration code for learning about AI/ML

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

The Modular Platform (includes MAX & Mojo)

LLM speculative inference server for consumer & heterogeneous hardware

LLM and Terminal inside VSCode for the Web

ELFIO and ARIO are robust, header-only C++ libraries designed to make binary file and archive manipulation effortless, efficient, and portable. Whether you are building compilers, linkers, binary a…

WebAssembly binding for llama.cpp - Enabling on-browser LLM inference

LLM inference in C/C++

ブラウザ上に実装された POSIX 準拠の最小 UNIX 互換 OS

An addon for xterm.js to add a PTY layer

Run a privacy-first AI coding agent directly in your browser with Python and local filesystem access.

P2P distributed LLM inference on browsers

chiwawa(CHeckpoint/restore and Instrumentation-specific WAsm runtime on WAsm runtime)

Kubernetes in Apple Containerization

WebAssembly (Wasm) arch support for the Linux kernel

Wassette: A security-oriented runtime that runs WebAssembly Components via MCP

Building a full-fledged code editor for iPad

A browser based code editor

High-performance In-browser LLM Inference Engine

Pyodide is a Python distribution for the browser and Node.js based on WebAssembly

Mount FUSE-based file systems without /dev/fuse

An open-source interactive visualization tool for comparing IR CFGs

social reputation checker for Go modules

A fast usermode x86 and x86-64 emulator for Arm64 Linux

Reverse engineering Rosetta 2 on M1 Mac

🍺 Alcoholless: lightweight security sandbox for Homebrew, AI agents, etc. on macOS

The fastest RISC-V sandbox

A very fast linker for Linux