Verilog in your browser — fully client-side. A browser-based Verilog simulator built by compiling Icarus Verilog to WebAssembly

macOS video editor built for AI

Modular AI HAL (Hardware Abstraction Layer) for Robots

ML kernels and benchmarking infrastructure written in TIRx

Build and run containers leveraging NVIDIA GPUs

Quick and Self-Contained AOTInductor Custom Operator Implementation and Integration

An Optimizer for Nvidia Compilers.

A CPU+GPU Profiling library that provides access to timeline traces and hardware performance counters.

Autonomous CUDA kernel optimization agent with structured task specs and per-config scoring

Machine learning compiler in Go. A from-scratch tinygrad port: graph-rewrite IR, autodiff as a compiler pass, zero-CGO WebGPU backend.

KVarN is a native vLLM KV-cache quantization backend for your agents: 3-5x more context, throughput above FP16, and FP16-level accuracy. Calibration-free, one flag.

Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models

Context Broker and CEF building block for context data management, providing NGSI interfaces.

Kafka based replacement for Cygnus

Agentic Kernel Optimization for All — automated GPU kernel optimization for any kernel, any hardware, any language

LibreYOLO is a MIT licensed open source computer vision library

Build your own high performance LLM inference engine in C++ and CUDA - a smaller version of vLLM

A GitHub action to install and setup a Linux distribution for the Windows Subsystem for Linux (WSL)

A SystemVerilog language server based on the Slang library.

A PyTorch native library for training speculative decoding models

KernelScript is a modern, type-safe, domain-specific programming language for eBPF-centric kernel customization

Fast and memory-efficient classical machine learning operators

The Triton backend for the ONNX Runtime.

CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs

Transformer with Local Modeling by Convolution for Speech Separation and Enhancement

llama-benchy - llama-bench style benchmarking tool for all backends

MiniJinja is a powerful but minimal dependency template engine for Rust compatible with Jinja/Jinja2

The GitHub Copilot app is an agent-native desktop experience for finding, running, steering, and landing software work across your GitHub repositories.

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