A Python framework for building distributed video and stream processing pipelines that can run locally or on Kubernetes using NATS JetStream.

Official SDKs for the Modal platform, enabling developers to deploy and high-performance serverless applications and interact with platform resources.

An open-source atomistic simulation engine built in PyTorch that accelerates Machine Learning Interatomic Potentials (MLIPs) through automatic batching and GPU acceleration.

An open-source implementation of a multimodal transformer model that natively processes text, images, and audio within a single architecture.

An auto-tuning tool for GPU kernels that benchmarks and optimizes parameters across multiple GPU programming languages to improve performance and and energy efficiency.

A toolbox of optimized, customizable building blocks for Transformer models that accelerates research by providing memory-efficient and high-performance CUDA kernels.

GenZ-ICP is a degeneracy-robust LiDAR odometry system that uses adaptive weighting to ensure stable localization in environments with few geometric features.

A modern multimodal recommendation toolbox that provides a unified framework and source code for a wide variety of research-based recommendation models.

A collection of agent skills and plugins that enable AI agents like Claude, ChatGPT, and Cursor to manage Railway cloud infrastructure.

A native macOS tool that uses AI agents to simulate real users navigating iOS, macOS, and web apps to identify UX friction and test user goals.

A memory adaptor for AI coding agents that provides a unified, cross-agent memory layer to store and recall project context and decisions.

A context engineering layer for AI agents that reduces token costs and improves accuracy by dynamically selecting only the most relevant tools and skills for each interaction.

A collection of 419 pre-trained AI models across various domains like vision, audio, and text, all runnable via a unified CLI with automatic weight downloads.

An open-source, local-first agentic IDE for long-running engineering work that preserves task history and environment state across sessions.

A highly extensible chatbot framework for Koishi that integrates multiple LLM providers and supports diverse output formats like voice and images.

An AI agent workspace for developers to run multiple parallel AI-powered workflows and automate development tasks from a single environment.

EdgeYOLO is an anchor-free object detector optimized for real-time performance on embedded edge devices like Nvidia Jetson, supporting multiple deployment formats including TensorRT and RKNN.

BrickGPT is an AI system that generates physically stable and buildable toy brick models from text prompts by fine-tuning a Llama-3.2 model.

OmniShow is a multimodal video generation model that unifies text, reference images, audio, and pose conditions to create high-fidelity videos of humans interacting with objects.

A high-performance codebase for experimenting with large diffusion models, implementing advanced flow-matching and distillation techniques for efficient few-step image generation.

A production-grade deployment toolkit for LLMs and VLMs based on PaddlePaddle, offering high-performance inference and broad hardware compatibility.

A TensorFlow library for building and scaling Graph Neural Networks, featuring tools for heterogeneous graph representation and distributed graph sampling.

PyKEEN is a Python package for training and evaluating knowledge graph embedding models, providing a comprehensive library of built-in datasets and models.

A visual data-flow environment for battle-testing and evaluating LLM prompts across multiple models and parameter permutations.

A native iOS and macOS workbench for testing and shipping with Apple's Foundation Models framework, featuring a prompt playground, tool integration, and adapter comparison.

A system that orchestrates multiple AI coding CLI instances in parallel using a feudal hierarchy and YAML-based communication to eliminate API coordination costs.

A collection of generalized, ready-to-use controllers for the ros2-control framework, compatible with MoveIt2 and Nav2.

An MCP server that enables AI agents to deploy and manage applications on Google Cloud Run through a standardized protocol.

An open-source, self-hostable Skills Hub and control plane for managing AI agent skills and deploying persistent AI teammates.

Paritok is an open‑source proxy that sits between coding agents (Claude Code, Cursor, Codex, OpenHands, etc.) and LLM APIs. It reduces input‑token usage by (1) filtering irrelevant tool schemas, (2) compressing file reads/tool outputs/history with a 4 B model, and (3) summarising old turns. The gateway is a Python package (`paritok[proxy]`) that can run locally (via Ollama or vLLM) or via a hosted GPU service. Reported savings are ~25 % on the first turn and >60 % in longer sessions, translating to noticeable cost reductions. The project is Apache‑2.0 licensed and includes a dashboard and VS Code extension.

A TypeScript framework for building AI workflows and agents that integrates prompt management, tracing, and evaluation into a single codebase to avoid SaaS fragmentation.

A minimalist, open-source LaTeX OCR model with only 20M parameters that converts mathematical formula images into LaTeX code.

Polyaxon is a platform for building, training, and monitoring large-scale deep learning applications, solving problems related to reproducibility, automation, and scalability.

ACI.dev is an open-source tool-calling platform that provides AI agents with unified, secure access to over 600 pre-built integrations via an MCP server or SDK.

A primitive parallel-formant speech synthesizer that recreates the retro sound of 1970s and 80s computer speech for the browser, Node.js, and CLI.

FDTDX is a JAX-based Python package for high-performance 3D electromagnetic simulations of photonic nanostructures, featuring automatic differentiation and GPU acceleration.

A GPU acceleration plugin for PySCF that speeds up quantum chemistry simulations, including Hartree-Fock and Density Functional Theory calculations.

A tool that uses LLMs to un-minify JavaScript by suggesting meaningful names for variables and functions while maintaining structural integrity via AST manipulation.

A browser control plugin for AI agents that enables direct Chrome navigation and interaction via the Chrome DevTools Protocol, available as a CLI tool or an MCP server.

A customizable, Rust-based statusline for Claude Code that displays real-time session cost, context usage, and API limits.

A suite of agentic reinforcement learning algorithms (ARPO and AEPO) that optimize LLM tool-use and reasoning by adaptively managing entropy during multi-turn interactions.

A universal benchmark for evaluating Audio Large Language Models (AudioLLMs) across 50+ datasets covering speech, audio-scene, and voice understanding.

An MCP server that gives AI agents the tools to interact with dbt projects, allowing them to execute SQL, analyze data lineage, and manage dbt pipelines.

An open-source adversary emulation framework for red-teaming AI agents and MCP servers, providing automated vulnerability scanning based on OWASP standards.

An agentic, cross-platform SQL database GUI client that supports 70+ databases and uses AI to generate, optimize, and fix SQL queries from natural language.

A Transformer-based SLAM system that unifies localization and mapping into a single model for 3D reconstruction and camera tracking from image sequences.

A Kubernetes operator that enables AI-driven cluster analysis and troubleshooting, featuring automated remediation for specific workload failures.

A model router that provides a single local endpoint to intelligently route AI requests between local, account-backed, and API-based models.