Libraries & frameworks

Zero-dependency front-end library providing beautiful, responsive, highly customizable, and accessible popup replacements for web pages with modern dialog components for developers.

The official PyTorch computer vision library, offering common datasets, image preprocessing transforms, and pretrained architectures for classification, detection, and related vision tasks.

A tiny JavaScript utility for conditionally joining CSS class names, widely used as a development dependency in React and other frontend projects handling dynamic styling logic.

A composable Python toolkit for creating command line interfaces with nesting and automatic help pages.

A high-performance, SEO-friendly lazy-loading library for the web that detects element visibility to defer loading of images, responsive images, iframes, and other resources.

Asynchronous SQL toolkit for Rust offering compile-time checked queries and connection pooling, supporting PostgreSQL, MySQL, and SQLite database operations for developers.

UI framework for building cross-platform mobile, desktop, and web applications in pure Python, using Flutter rendering with rich controls and packaging deployment support.

High-performance compiler translating Python syntax into native machine code, with built-in NumPy support, multithreading, and GPU capabilities for substantial speedups.

React rich-text editor framework built on Slate, offering plugin system and shadcn components to help developers quickly build customizable rich-text editors.

A TypeScript library for building robust, type-safe production applications with typed errors, dependency injection, and structured concurrency.

Lightweight deep learning framework from Alibaba for mobile and edge developers, supporting efficient on-device inference and training, including deployment of large language and diffusion models.

Rust full-text search engine library inspired by Lucene, providing developers with indexing, tokenization, and retrieval primitives for building custom search applications.

Python library that converts pandas DataFrames into drag-and-drop interactive visualizations, helping data analysts explore, clean, and analyze tabular data directly inside Jupyter notebooks.

Next-generation tensor library and deep learning framework written in Rust, giving developers flexible and portable tools for both model training and inference with high efficiency.

A development library for guaranteed structured outputs from large language models, letting the same code enforce type-valid generation across multiple model backends.

ModelScope's fine-tuning and deployment framework supporting 600+ LLMs and 300+ multimodal models for pre-training, fine-tuning, alignment, quantization, inference, evaluation, and deployment.

Generate massive amounts of fake data in the browser and node.js

Open-source ETL library for LLM applications that ingests and preprocesses documents, offering partitioning, cleaning, chunking, and embedding for multi-format RAG pipelines.

Open-source face recognition library based on deep neural networks, providing researchers with tools for model training, facial feature representation, and face verification.

Portable C tensor library for machine learning providing efficient computation and quantization kernels used as a foundation for on-device and cross-platform model runtimes.

Zero-dependency universal JavaScript library for RFC-compliant UUID generation and parsing, supporting developers with multiple UUID versions across all Node and browsers.

The API and real-time application framework

Offline open-source speech recognition toolkit supporting over 20 languages, giving developers multi-language API bindings for chatbots, smart homes, subtitling, and transcription.

CUDA-compatible replacement layer for non-NVIDIA GPUs that runs unmodified CUDA applications with near-native performance as a general GPU compute runtime library.

C++ machine learning and data analysis toolkit providing algorithms for classification, regression, clustering, face detection, and more for building complex software applications.

A delightful Python library that turns OpenStreetMap data into beautiful custom maps with presets and a Streamlit front end.

Open-source implementation library for CLIP models providing training and inference code, multi-architecture support, and pretrained weights for large-scale image-text contrastive learning research.

A pure PHP library for reading and writing spreadsheet files like Excel and ODS in any PHP project.

Python library adding reliable structured JSON output to various large language models. Built on Pydantic, it provides validation, retries, and type safety for application developers.

Open machine learning compiler framework translating models into optimized modules for diverse hardware targets. Lets engineers customize compilation pipelines through Python for efficient deployment.

Data quality profiling and exploratory analysis library requiring one line of code for Pandas and Spark DataFrames, generating HTML, JSON, and Jupyter reports for analysts.

Developer library providing clean from-scratch implementations of 20+ large language models. Supports pretraining, fine-tuning, quantization, inference, and recipes for large-scale deployment.

Simple peer-to-peer with WebRTC.

Compact general-purpose memory allocator written in C as a drop-in malloc replacement. Offers strong performance across platforms for systems programmers building performance-sensitive applications.

NVIDIA SDK for high-performance deep learning inference on GPUs. Optimizes trained models for deployment with plugins, an ONNX parser, and example applications for developers.

Supplies Meta's open-source, customizable, agent-ready design system with accessible components, theming, and CLI tooling, letting developers and agents build interfaces consistently.

TypeScript compiler and toolchain for building framework-agnostic Web Components. Enables teams to maintain one component codebase and distribute enterprise design systems across frontend frameworks.

Open project improving LLM function calling through fine-tuning, datasets, and benchmarks. Publishes the Gorilla model family and BFCL evaluation suite for researchers and developers.

A collection of JavaScript libraries for Progressive Web Apps handling service workers, caching, and offline experiences.

Python library bringing diverse CAM interpretability methods to PyTorch vision models. Visualizes predictions with heatmaps for classification, detection, and segmentation to aid diagnosis.

Easy-to-use development library for large language models built on PaddlePaddle. Provides a broad model collection with training, fine-tuning, and inference capabilities for NLP tasks.

Monorepo containing Grafast, PostGraphile, and related PostgreSQL tooling. Automatically generates high-performance GraphQL APIs and SQL builders for backend developers working with Postgres.

Unified codebase and benchmark suite for deep time-series analysis. Covers long- and short-term forecasting, imputation, anomaly detection, and classification for researchers.

Open-source LLM fine-tuning framework supporting diverse models and parameter-efficient methods, offering configuration-driven training workflows for machine learning practitioners to streamline experimentation and production adaptation.

Fast, lightweight open-source document parser extracting text, layout, and bounding boxes locally, with support for multiple languages, formats, and downstream processing outputs.

GPU-accelerated array-computing library for scientists and engineers, mirroring NumPy and SciPy interfaces while exploiting NVIDIA and AMD hardware for high-throughput numerical workloads.

Modular Firebase client library for React Native developers, bridging native iOS and Android SDKs to provide authentication, databases, messaging, analytics, and other full Firebase services in JavaScript.

Comprehensive retrieval toolkit offering BGE embedding and reranking models with training and inference code, supporting search and RAG application development for NLP practitioners.

Free React map component library for frontend developers, built on MapLibre and Tailwind with zero-config setup, providing composable and customizable UI blocks for interactive web maps.

React component and block library built on shadcn/ui, offering customizable production-ready UI for Next.js developers with quick CLI-based integration to accelerate development.

Development library bringing CSS-like style definitions to Go terminal applications, supporting colors, typography, alignment, and layout rendering for command-line interfaces.

PyTorch-based open-source speech processing toolkit providing training recipes and pretrained models for diverse speech and text tasks to accelerate conversational AI development.

HTTP extension module embedding Lua into NGINX, providing non-blocking APIs and configuration directives that let developers handle requests in Lua as a core OpenResty component.

Declarative deep learning framework that trains, fine-tunes, and deploys LLMs, tabular, and multimodal neural networks via YAML configuration without boilerplate code.

Open-source JavaScript library for rendering GPU-accelerated interactive vector maps in browsers, enabling web developers to embed customizable zoomable maps into websites and applications.

WeChat's cross-platform mobile database framework on SQLite offering ORM, encryption and repair for C++, Java, Kotlin and Swift.

A tiny JavaScript debugging utility modelled after Node.js core's debugging technique. Works in Node.js and web browsers

🛏 An HTML to Markdown converter written in JavaScript

Differentiable computer vision library built on PyTorch offering 500+ operators for image processing, augmentation, and geometric vision with batch transforms, autograd, and GPU acceleration.

Minimal educational implementations of a dozen classic machine learning algorithms using NumPy, helping learners study principles and practice building models from scratch.

Provides open-source speech-to-text, intent detection, and text-to-speech models and toolkits optimized for ultra-low latency on-device inference, enabling developers to build real-time voice agents across multiple platforms.

Socket-based TypeScript and JavaScript API for WhatsApp Web, allowing developers to send and receive messages and manage sessions and contacts in custom applications.

Open-source toolkit for optimizing and deploying AI models from multiple frameworks, converting vision, speech, and generative models to run efficiently on CPU, GPU, and NPU hardware.

Python AutoML library enabling modeling on tabular, time-series, and multimodal data with minimal code, handling training, tuning, and prediction for analysts and applied machine learning practitioners.

Webpack build plugin that automatically generates HTML files and injects bundled assets, supporting custom templates and hashed filenames for front-end engineers simplifying project configuration.

PyTorch implementation of denoising diffusion probabilistic models providing U-Net architectures, diffusion processes, and trainers for researchers training and sampling image and sequence generation models.

Ultralytics implementation of YOLOv3 object detection providing training, validation, inference, and multi-format export utilities for developers working with three classic detection models.

General-purpose Python library for building interactive command-line applications, offering syntax highlighting, autocompletion, multiline editing, and Emacs/Vi key bindings for CLI developers.

Lightweight JavaScript search engine library running in browsers, servers, and edge environments, supporting full-text, vector, and hybrid retrieval plus RAG for site-search developers.

NVIDIA's high-performance CUDA C++ template library providing reusable GEMM and linear algebra components plus Python DSL for researchers and engineers writing optimized GPU kernels.

Research-oriented reinforcement learning library offering high-quality single-file implementations of algorithms like PPO, DQN, and SAC, with benchmarking and large-scale experiment support.

An open-source C++ toolkit for creative coding, offering cross-platform libraries for graphics, video, audio, and computer vision projects.

Distributed drop-in replacement for pandas that accelerates large-scale dataframe processing across multicore CPUs using Ray or Dask engines with a single-line code change.

A comprehensive React UI library and design system with quality components and design-to-code tooling.

.NET (WPF and Windows Forms) bindings for the Chromium Embedded Framework

PHP extension client for Redis, Valkey, and compatible key-value stores, wrapping connections and commands so PHP developers can interact with caches and data structures in code.

React animation library managing mount and unmount states as components enter or leave the DOM, helping front-end developers implement structured enter-exit transition effects.

Lightweight dependency-free JavaScript library for parsing User-Agent strings to identify browsers, operating systems, devices, crawlers, and AI bots for analytics and compatibility logic.

Home of the WebKit project, the browser engine used by Safari, Mail, App Store and many other applications on macOS, iOS and Linux.

Build native cross-platform mobile, web, desktop and embedded apps from a single C#/XAML codebase using open-source .NET UI framework for developers targeting multiple platforms efficiently.

Privacy-preserving Python library for data science where data stays with owners, allowing analysts to run only approved computations remotely and receive results without direct access.

Comprehensive Python anomaly detection library with 60+ detectors for tabular, time-series, graph, text, image, and audio data, offering unified APIs and benchmarks for practitioners and researchers.

TensorFlow 2-based neural network library offering composable Module abstraction and common layers for machine learning researchers building and experimenting with models.

PyTorch-based end-to-end speech processing toolkit covering recognition, synthesis, translation, and enhancement, with reproducible recipes and pretrained models for speech researchers and engineers.

A widely used PHP library component that simplifies building beautiful, testable command-line interfaces with commands, arguments, and styling helpers.

Minimal high-performance Python web API and microservices framework supporting WSGI/ASGI, helping backend developers build reliable and scalable REST APIs efficiently.

A BPMN 2.0 rendering toolkit and web modeler.

Drop-in enhancement library for Node.js native fs module, adding copy, remove, and mkdirs operations with Promise support for JavaScript developers handling files.

A cross-platform UI automation framework for games and apps that uses image recognition to script tests once and run anywhere.

Web scraping and browser automation library for Python developers supporting HTTP and Playwright crawling, proxy rotation, persistent storage, ideal for extracting web data for AI workflows.

End-to-end open-source library for LLM/VLM development supporting fine-tuning, reinforcement learning, evaluation, data synthesis, and deployment for AI teams building custom models.

High-performance C++ asynchronous server framework based on futures for writing non-blocking applications, adopted by systems like ScyllaDB for backend developers.

An open-source JavaScript engine with a C API for embedding JavaScript support into C and C++ applications.

A highly configurable Node.js logging library offering beautiful out-of-the-box loggers, timers, scopes, and secrets filtering for developer consoles.

A Python library providing friendlier date and time handling, including creation, parsing, formatting, timezone conversion, timespan generation, and humanized display for developers.

A C++ foundation library offering the high-performance multithreaded TCMalloc allocator alongside heap and CPU profiling tools for performance analysis and optimization.

A reusable Go component library for Bubble Tea apps, offering inputs, tables, spinners, progress bars, paginators and pickers.

A framework letting developers build custom Windows file systems in user mode without kernel programming.

A lightweight JavaScript runtime built on Rust and QuickJS, optimized for AWS Lambda serverless environments with fast startup time and minimal memory footprint.

Meta-programming framework exploring spatiotemporal composability theory and implementation. Targeted at programming-language researchers, its API remains unstable and under active development for composition experiments.