Technical intelligence for agents and analysts

Model definition, loading, and fine-tuning framework for transformer-based architectures across text, vision, audio, and multimodal domains with unified APIs for inference and training.

stars: 159,006

Official NVIDIA high-performance inference optimization library for Large Language Models on NVIDIA hardware, providing advanced kernels, quantization, and orchestration.

stars: 13,319

High-throughput, memory-efficient LLM inference and serving engine with optimized batching, KV-cache management, and multi-GPU/hardware support

stars: 75,657

Meta-operating system and middleware for robot development, providing communication, coordination, and tooling across distributed robot systems

stars: 5,312

End-to-end open-source software stack for autonomous driving, providing modules for sensing, localization, perception, planning, and control.

stars: 11,344

Open-source autopilot flight control software stack for multirotor, fixed-wing, and VTOL unmanned aerial vehicles (UAVs), providing real-time control algorithms, sensor fusion, navigation, and hardware abstraction across diverse embedded platforms.

stars: 11,461

Production-grade inference serving platform for deploying and managing machine learning models across cloud and edge environments with multi-backend support, batching, and dynamic loading.

stars: 10,527

LLM application framework providing abstractions for chaining language models, memory, retrieval, and agents with pluggable integrations across 100+ external services

stars: 132,702

Official Go implementation of the Model Context Protocol (MCP) for building interoperable AI clients and servers that connect LLMs to data sources and tools.

stars: 4,316

Model Context Protocol: A standard specification and reference implementation for connecting AI models to external data sources, tools, and context via a unified protocol

stars: 7,745

Production-grade WebAssembly runtime with JIT compilation, sandboxing, and WASI support for executing WASM modules across multiple platforms

stars: 17,866

Workflow orchestration platform for authoring, scheduling, and monitoring data pipelines and ETL processes

stars: 44,937

Framework for building AI agents with structured outputs, tool use, and multi-model support via Pydantic validation

stars: 16,164

Open lakehouse data format optimized for multimodal AI workloads, providing fast random access, vector indexing, and data versioning with seamless integration across data science ecosystems.

stars: 6,282

Long-horizon agent orchestration framework enabling multi-step reasoning, code generation, and task execution through sandboxed environments, persistent memory, tool integration, and hierarchical subagent coordination.

stars: 57,861

Graph-based Retrieval-Augmented Generation (RAG) system that extracts entities and relationships from documents to build knowledge graphs for improved LLM context retrieval

stars: 32,039

Unified LLM API abstraction layer and gateway that normalizes requests/responses across 100+ LLM providers into OpenAI-compatible format, with cost tracking, load balancing, and guardrails.

stars: 42,519

Official C# SDK for implementing Model Context Protocol (MCP) servers and clients, enabling .NET applications to interact with LLM-powered services via a standardized protocol.

stars: 4,174

Production-grade ROS 2 navigation framework providing autonomous robot path planning, localization, and costmap-based obstacle avoidance for mobile robotics

stars: 4,113

Orchestration framework for building stateful, multi-agent applications using graph-based logic with support for cycles and persistence.

stars: 2,755