AI Data Tools

High-performance S3-compatible object storage

MinIO is a high-performance, S3-compatible object storage server designed for AI, machine learning, and data-intensive workloads. Written in Go, it delivers industry-leading throughput for both read and write operations while maintaining full compatibility with the Amazon S3 API. MinIO includes an embedded web console for bucket management, a command-line client, and supports erasure coding, bitrot protection, and encryption at rest for enterprise-grade data durability.

High-performance mobile neural network inference

NCNN is Tencent's high-performance neural network inference framework optimized for mobile and embedded platforms. It features pure C++ with zero dependencies, ARM NEON assembly optimization, Vulkan GPU acceleration, and sophisticated memory management for minimal footprint. Supports importing models from PyTorch, ONNX, Caffe, TensorFlow, and Keras with 8-bit quantization and half-precision storage for efficient on-device deployment across Android, iOS, and Linux.

Private, self-hosted research notebooks with flexible AI models, source chat, and podcasts

Open Notebook is an MIT-licensed, self-hosted alternative to NotebookLM for collecting sources, chatting over research, generating reusable transformations, and producing multi-speaker podcasts. Its Docker stack keeps notebook data under the user's control while supporting 18-plus model providers, including local Ollama and LM Studio workflows.

AI-ready PDF parser with benchmark-leading accuracy

OpenDataLoader PDF is a high-performance parser that extracts structured, AI-ready data from PDFs with industry-leading 0.907 benchmark accuracy. Combines deterministic local processing with optional AI hybrid mode for complex layouts, OCR support across 80+ languages, formula extraction in LaTeX, chart descriptions, and built-in prompt injection filtering. Available as Python, Node.js, and Java SDKs for seamless RAG pipeline and data preparation integration.

End-to-end open-source platform for training and evaluating foundation models

Oumi is an end-to-end open-source platform for training, fine-tuning, and evaluating foundation models at any scale. It covers data preparation, distributed training, reinforcement learning from human feedback, evaluation benchmarks, and model deployment in a unified framework. Supports training from scratch to post-training alignment with over 9,100 GitHub stars.

Data versioning and pipeline automation for ML

Pachyderm is a data versioning and pipeline automation platform that provides Git-like version control for datasets with automatic data lineage tracking. Acquired by HPE, it enables reproducible ML workflows by connecting data versioning to containerized processing pipelines. Features include automatic provenance tracking, incremental processing, and deduplication for efficient storage of large datasets.

State-of-the-art OCR toolkit supporting 100+ languages from Baidu

PaddleOCR is an open-source OCR toolkit from Baidu's PaddlePaddle ecosystem with over 73,000 GitHub stars. It provides ultra-lightweight and high-accuracy text detection and recognition for 100+ languages including CJK, Arabic, and Indic scripts. The toolkit offers pre-trained models, easy deployment via pip, and server/edge inference options for document digitization workflows.

Real-time ETL and RAG engine with Python API and Rust core

Pathway is an open-source Python ETL framework with a high-performance Rust engine for stream processing, real-time analytics, and RAG pipelines. It handles both batch and streaming data in a unified API, enabling live-updating vector indexes, real-time document processing, and AI agent memory that refreshes as new data arrives. 63,000+ GitHub stars, used for production RAG systems where static vector databases create stale context. Apache 2.0 licensed with enterprise cloud options.

Declarative multimodal AI data infrastructure

Pixeltable is a declarative data infrastructure for multimodal AI that stores video, audio, images, and documents as first-class column types. Define Python computed columns for inference and transformations, and Pixeltable auto-orchestrates execution with incremental updates. Built-in vector search eliminates the need for separate vector databases while supporting RAG and semantic search workflows.

Lightning-fast DataFrame library in Rust

Polars is an extremely fast DataFrame library written in Rust that provides a powerful query engine for data manipulation in Python, Node.js, and R. Built on Apache Arrow columnar format, Polars delivers performance that outpaces Pandas by 10-100x on common operations through parallel execution and SIMD optimizations. It features lazy evaluation with automatic query optimization, streaming for out-of-core processing, and an expressive API for filtering, joining, and aggregating datasets.

Open-source PII detection and anonymization for AI data flows

Presidio is an MIT-licensed privacy framework for identifying and anonymizing personally identifiable information in text, images, and structured data. It can act as a de-identification layer around LLM prompts, logs, RAG corpora, and customer-data workflows.

First commercially viable 1-bit LLMs that are 14x smaller and 8x faster

PrismML Bonsai delivers the first commercially viable 1-bit large language models with 8B, 4B, and 1.7B parameter variants. The 8B model runs in just 1GB of RAM versus 16GB for standard FP16 models, achieving 44 tokens per second on iPhone. Backed by $16.25M from Khosla Ventures and released under Apache 2.0, Bonsai makes capable LLMs practical for edge devices and resource-constrained environments.

Optimize and deploy AI models on Snapdragon devices

Qualcomm AI Hub is a platform for optimizing and deploying AI models on Snapdragon-powered devices with NPU acceleration. It provides pre-optimized models, profiling tools, and the SNPE SDK for compiling models to run efficiently on Qualcomm's Hexagon DSP and AI Engine. Supports hundreds of model architectures with on-device benchmarking across real Snapdragon chipsets for mobile, IoT, and XR applications.

All-in-one multimodal RAG framework

RAG-Anything is an all-in-one multimodal RAG framework from the University of Hong Kong that processes text, images, tables, and equations through a unified pipeline built on LightRAG. It constructs multi-modal knowledge graphs by extracting multimodal entities and establishing cross-modal relationships. The VLM-Enhanced Query mode integrates visual content into large language models for deeper document understanding beyond plain text retrieval.

On-device AI inference for React Native apps

Declarative framework for running AI models on-device in React Native applications, powered by Meta ExecuTorch runtime. Supports LLMs including Llama 3.2, computer vision, OCR, embeddings, and vision-language models on iOS 17+ and Android 13+. Developed by Software Mansion with pre-built optimized models, custom model export support, and privacy-first inference without any cloud dependency for mobile AI development.

Cross-platform on-device AI inference SDK

RunAnywhere SDK is a production-ready toolkit for running AI models entirely on-device across iOS, macOS, Android, Web, React Native, and Flutter. It provides a unified C++ core with platform-specific bindings for LLM text generation via llama.cpp, vision-language models, Whisper speech-to-text, Piper text-to-speech, and on-device image generation. All processing stays local with zero cloud dependency, ensuring privacy and low latency for mobile and edge AI applications.

AI-native state store with hybrid vector and full-text search

SeekDB is an open-source AI-native state store from the OceanBase ecosystem that combines MySQL-compatible data access with hybrid vector and full-text retrieval. It targets agent and AI application teams that need embedded or server deployment, copy-on-write style sandboxes, and searchable state without gluing together several separate storage layers.

Data-centric AI platform for programmatic data labeling

Snorkel AI is a data-centric AI platform that enables programmatic labeling of training data through labeling functions rather than manual annotation. Spun out of Stanford AI Lab, it lets teams write Python functions that encode domain heuristics to label data at scale, with the platform combining weak labels into high-quality training sets. Used by Fortune 500 companies for text, image, and structured data labeling.

Reusable computer vision tools for developers

Supervision is an open-source Python toolkit by Roboflow providing reusable CV utilities for detection, tracking, annotation, and dataset management. It works with any model including YOLO and Hugging Face via a standardized Detections class. Features include 20+ annotators, ByteTrack object tracking, zone counting, speed estimation, and dataset conversion between COCO, YOLO, and Pascal VOC formats.

Open-source library for generating synthetic tabular data

Synthetic Data Vault (SDV) is an MIT-backed open-source Python library for generating synthetic tabular, relational, and time-series data. It learns statistical patterns from real datasets and produces synthetic versions that preserve distributions, correlations, and referential integrity. Supports single-table, multi-table, and sequential data with built-in privacy and quality metrics.

Google's lightweight ML framework for mobile and embedded

TensorFlow Lite is Google's lightweight ML framework for deploying models on mobile and embedded devices. It supports quantization, GPU/NPU delegation, and runs on Android, iOS, Linux, and microcontrollers. Provides pre-trained models, model conversion tools from TensorFlow and JAX, and hardware acceleration via GPU, Hexagon DSP, and CoreML delegates. Powers on-device ML in billions of Google app installations.

Hugging Face's open-source inference server for embeddings, rerankers, and classifiers

Text Embeddings Inference is Hugging Face's Apache-2.0 server for high-throughput embedding, reranking, and sequence-classification models. TEI packages token-based dynamic batching, optimized Transformers kernels, Safetensors loading, OpenAI-compatible embedding endpoints, Prometheus metrics, and configurable OpenTelemetry tracing in deployable CPU and GPU images.

Google's pretrained foundation model for zero-shot time-series forecasting

TimesFM is a pretrained time-series foundation model from Google Research that performs zero-shot forecasting on diverse datasets without task-specific training. It handles univariate and multivariate time series across domains including finance, logistics, energy, and infrastructure monitoring with accuracy competitive against traditional statistical methods like ARIMA and Prophet.

Offline speech recognition for 20+ languages

Vosk is an offline speech recognition toolkit supporting 20+ languages with compact 50MB models that run on Raspberry Pi, Android, iOS, and servers. It provides streaming API with zero-latency response, speaker identification, and reconfigurable vocabulary. Vosk offers bindings for Python, Java, Node.js, C#, Go, and Rust. Unlike cloud-based alternatives, all processing happens locally with no internet required. Apache 2.0 licensed with 14K+ GitHub stars.

Tokenizer-free multilingual TTS with voice cloning

VoxCPM is an open-source text-to-speech system from OpenBMB generating continuous speech across 30 languages without traditional tokenization. Its 2B parameter end-to-end diffusion architecture produces 48kHz studio-quality audio with natural prosody and emotion. Key capabilities include voice design from text descriptions, few-shot voice cloning, and multilingual synthesis without language-specific modules. The Apache 2.0 project has 8,700 GitHub stars.

Enterprise RAG framework by Tencent

WeKnora is a Tencent-developed LLM-powered knowledge management and Q&A framework for enterprise document understanding and semantic retrieval. Supports 10+ document formats including PDF, Word, Excel, and images with seamless IM platform integration for WeCom, Feishu, Slack, and Telegram. Offers Quick Q&A mode using RAG pipelines and Intelligent Reasoning mode with ReACT agents for complex multi-step reasoning tasks across organizational knowledge bases.

OpenAI's open-source speech recognition model for any language

Whisper is OpenAI's open-source automatic speech recognition model trained on 680,000 hours of multilingual audio data. It supports transcription and translation across 99 languages with robust handling of accents, background noise, and technical vocabulary. Available in multiple model sizes from tiny (39M) to large (1.5B parameters) for balancing accuracy and speed.

Context engineering platform for AI agents with temporal knowledge graphs

Zep is a context engineering platform that assembles relationship-aware context for AI agents from conversations, business data, documents, and events. It maintains a temporal knowledge graph that automatically extracts entities and relationships, tracking how context evolves over time. Zep delivers formatted context blocks optimized for LLMs with sub-200ms latency, integrating with LangChain, LlamaIndex, AutoGen, and Google ADK through Python, TypeScript, and Go SDKs.

Lightweight C++ inference for Google Gemma models

gemma.cpp is Google's standalone C++ inference engine built specifically for running Gemma language models without Python or CUDA dependencies. It provides optimized CPU inference using SIMD instructions and Highway library, supports Gemma 2 and Gemma 3 models, and runs on x86 and ARM architectures. Designed for embedded systems, edge devices, and server deployments needing minimal overhead.

Git-like version control for data lakes and object storage

lakeFS is an open-source platform that brings Git-like branching, committing, and merging to data lakes and object storage. It works on top of S3, GCS, Azure Blob, and MinIO, enabling teams to create isolated data branches for experimentation, run CI/CD for data pipelines, and maintain full data lineage. Acquired DVC in 2025, uniting data version control for both small and enterprise-scale workloads.

ModelScope's fine-tuning framework supporting 600+ models

ms-swift is ModelScope's open-source framework for fine-tuning over 600 large language and multimodal models. It supports SFT, DPO, RLHF, LoRA, QLoRA, and full fine-tuning with a web UI and CLI interface. Optimized for the Chinese AI ecosystem with native ModelScope Hub integration alongside Hugging Face support. Over 13,500 GitHub stars.

Meta's official PyTorch library for LLM fine-tuning

torchtune is Meta's official PyTorch-native library for fine-tuning large language models. It provides composable building blocks for training recipes covering LoRA, QLoRA, full fine-tuning, DPO, and knowledge distillation. Supports Llama, Mistral, Gemma, Qwen, and Phi model families with distributed training across multiple GPUs. Designed as a hackable, dependency-minimal alternative to higher-level frameworks.

Production-grade reinforcement learning framework for LLM training

verl is an open-source reinforcement learning framework designed specifically for training and aligning large language models. Built for production use with support for distributed training across multiple GPUs and nodes, it implements RLHF, DPO, and other alignment algorithms that make LLMs follow instructions, avoid harmful outputs, and generate higher quality responses. Over 580 contributors and 20,000 GitHub stars signal strong adoption.