Trending AI projects

FAROS is a multi-agent AI system designed to automate scientific research workflows, from literature review and experiment planning to evidence auditing and paper writing, while ensuring all findings are traceable and verifiable.

A comprehensive Python library for Thai Natural Language Processing, providing tools for tokenization, POS tagging, and linguistic analysis.

Gloss is an open‑source web app that lets you ask questions about any selected line in an article or PDF and get AI‑generated, cited answers right in the margin, using Together.ai LLMs, Exa search, and client‑side PDF parsing.

A high-performance TopK kernel implementation for DeepSeek Sparse Attention and sampling, offering 2-20x speedup over torch.topk on CUDA and Ascend platforms.

A real-time webcam application that detects facial expressions and hand gestures to trigger matching meme overlays, featuring a virtual camera for use in video calls.

An AI-powered toolkit for e-commerce sellers to automatically analyze product photos and generate a complete set of professional marketing images across various styles and platforms.

A proxy server that exposes an Anthropic-compatible API backed by Antigravity's Cloud Code, enabling the use of Claude and Gemini models with the Claude Code CLI.

OlmoEarth is an open‑source library for building and using multimodal, spatio‑temporal foundation models on satellite and derived Earth‑observation data. It ships pretrained checkpoints (Nano‑to‑Large) trained on Sentinel‑1/2, Landsat, and six map layers, and includes scripts for pre‑training, fine‑tuning, inference, and evaluation.

WebODM is commercial-grade software for processing drone images into georeferenced maps, 3D models, point clouds, and elevation models.

A synthetic patient population simulator that generates realistic but non-real health records in various formats like FHIR and CSV for testing and development.

An agent-first business intelligence tool that uses BI-as-code and OLAP engines to provide real-time analytics and conversational BI for humans and AI agents.

A comprehensive, reproducible wiki for serving frontier LLMs on NVIDIA RTX PRO 6000 Blackwell GPUs, offering Docker images, exact launch scripts, benchmarks, quantisation fidelity checks, and low‑level hardware‑tuning guidance.

An OpenAI-compatible API gateway for Jimeng AI, providing access to its image and video generation models with built-in account pooling and risk-control bypass.

A self-hosted AI system that watches 3D printers via camera, detects failed prints in real-time using a compact on-device vision model, and automatically pauses/cancels the print and alerts you — no cloud required.

A comprehensive digital human framework that integrates LLMs, ASR, and TTS to power interactive virtual avatars for apps, websites, and embedded devices.

A topic modeling technique that leverages transformers and c-TF-IDF to create dense, interpretable clusters of documents.

A PyTorch-based open-source framework for deep learning in healthcare imaging that provides standardized workflows and domain-specific tools for medical data.

Datus is an open‑source Python agent that uses LLMs to automate data‑engineering tasks—semantic modeling, SQL generation, ETL pipeline creation, and report/dashboard building—while storing each interaction in a context engine that improves accuracy over time. It offers a CLI, web chat, and API, integrates with many databases, LLM providers, schedulers and BI tools, and is extensible via a plug‑in system.

minutes is an open‑source desktop/CLI app that records, locally transcribes, and stores meetings as Markdown files with rich metadata. It exposes a simple MCP API (`search_meetings`) so any LLM agent (Claude Code, Cursor, Codex, etc.) can retrieve relevant excerpts, enabling privacy‑first, agent‑first workflows.

A neural-network powered karaoke system that converts any song in a user's music library into a karaoke track by separating vocals and generating synchronized lyrics.

A GPU-accelerated elevation mapping framework for legged robots that fuses multi-modal data like geometry and semantics to create accurate real-time terrain maps.

Earth2Studio is a Python-based toolkit that provides a unified interface for building and running AI-driven weather and climate forecasting pipelines, allowing users to easily swap between different models and data sources.

PDD Alpha is a Node.js CLI that uses Codex or Claude Code to turn a GitHub issue into a real pull request via staged agent runs, with manual approval gates and your own test execution. It’s alpha software with many caveats, but the core flow is: read issue → plan → you approve → agents code → PDD tests → push PR → you merge.

OpenAI4S is a hybrid research agent that combines structured JSON orchestration with persistent Python/R kernels to automate scientific workflows, data analysis, and simulations.

BoltzGen is an open-source pipeline that designs novel protein binders for given target proteins, peptides, or small molecules. It uses diffusion-based backbone generation, inverse folding, and structure-based filtering to produce a ranked, diverse set of high-quality binder candidates.

An open-source workflow that uses natural language to automate the creation of book promotional short videos, covering everything from book selection to final rendering.

MeiGen AI Design MCP is a toolset that enables AI assistants to perform image and video generation, inspiration searching, and asset management directly within the chat interface using the Model Context Protocol.

A skill-centric agent runtime that allows AI agents to design, execute, and persist reusable skills to improve their performance on complex tasks through a reflection-based learning loop.

A Python library for automating the recognition and solving of various CAPTCHA types, including OCR, sliders, and rotation puzzles.

An OCR tool specifically optimized for Japanese manga, capable of recognizing multi-line text, furigana, and both vertical and horizontal layouts.

A deployment recipe for serving the Qwen3.8-Flash-Next vision-language model on a single DGX Spark using vLLM, with memory optimizations, speculative decoding, and extensive benchmark data.

A curated color-science knowledge base packaged as an agent skill, giving coding agents expert-level guidance on color theory, tools, and techniques. It solves the problem of agents lacking deep color knowledge by providing a layered, indexed collection of references, along with evaluation prompts to verify quality.

Pi‑intercom is a Pi‑extension that adds a local broker and a `/intercom` UI/tool, letting multiple Pi AI‑agent sessions on the same machine send, ask, reply, hand‑off, and cancel messages one‑to‑one. It supports aliases, blocking asks, safety knobs, and integrates with pi‑subagents for supervisor escalation, enabling coordinated multi‑agent workflows like planner‑worker delegation.

AVTR‑1 is a TensorRT‑accelerated, flow‑matching autoregressive model that turns a portrait image and dual‑audio streams into a 25 fps lip‑synced talking avatar. The repo includes weights, build scripts for GPU‑specific inference engines, an interactive web demo, and a CLI for offline generation. Real‑time performance is shown for several NVIDIA GPUs. Licensing is split: a community license for the model (commercial use under $10 M revenue), and non‑commercial PolyForm licenses for the renderer and streamer, with additional research‑only restrictions from InsightFace.

Big-AGI is a high-performance, multi-model AI workspace that helps experts reduce hallucinations and improve output quality by comparing responses across different AI services in a single, unified interface.

PaiAgent is an open‑source, enterprise‑grade platform that lets users visually compose AI workflows with drag‑and‑drop nodes. It supports multiple large‑model providers via Spring AI, offers two execution engines (a classic DAG engine and a LangGraph4j state‑graph engine), includes a declarative “Skills” system for LLM‑driven function calls, and provides real‑time SSE debugging. The stack is React + TypeScript on the front‑end and Spring Boot 3 + Java 21 on the back‑end, with MySQL (and optional MinIO) for persistence.

An MCP server that enables AI agents to use radare2 for binary analysis and reverse engineering.

A reproducible benchmark suite that runs LLMs (both local Ollama models and cloud providers) through a two‑phase autonomous coding task: generate a small Rails app, then build and validate it with Docker. It provides scripts for warm‑up, execution, and runtime validation, plus extensive markdown reports and hand‑written code reviews that score models on a 0‑100 rubric across eight quality dimensions.

OpenSplat is a C++ implementation of 3‑D Gaussian splatting that converts COLMAP/OpenSfM/ODX/nerfstudio data into compact Gaussian‑based scene files. It supports CUDA, ROCm (HIP), Apple Metal, and CPU‑only builds, offers Docker images, and provides a simple CLI for training, resuming, masking, and exporting to .ply/.splat/.spz/.rad formats.

A terminal-native, AI-powered image search tool that lets you search local folders using natural language or example images without needing to upload files.

FlyPython is an open‑source, bilingual repository that teaches Python through AI‑assisted challenge courses. You pick a course, let a coding agent fetch and run the files, and verify your solution with `verify.py`, which issues claim codes for each passed checkpoint. The repo also provides guides, a curated catalog of Python resources, and machine‑readable manifests for tool integration.

GeoAgent is a shared AI agent layer for geospatial Python libraries, map widgets, and QGIS plugins. It provides a consistent interface for LLMs to use geospatial tools, supports many AI providers, and includes safety confirmations. It offers factories for leafmap, anymap, QGIS, NASA OPERA, and STAC, plus a QGIS plugin (OpenGeoAgent) with a chat UI.

MakeACopy is an offline-first Android document scanner that uses local machine learning models for edge detection and OCR to create searchable PDFs without any cloud dependency.

A macOS sandboxing tool that uses sandbox-exec to restrict LLM coding agents to a least-privilege file system access model.

A collection of opinionated configurations, sandboxing rules, and hooks for Claude Code to enable secure, high-throughput autonomous AI coding.

A cross-platform desktop application that provides a unified, visual interface to manage configurations, sessions, and plugins for various AI coding assistants and CLI tools.

Bug Hunter is an adversarial AI code-audit tool that uses a multi-agent system of hunters, skeptics, and referees to filter out false positives and provide evidence-backed security and behavioral bug reports.

A zero-dependency CLI tool that lets you generate images using your existing ChatGPT or Gemini subscriptions without requiring separate API keys or additional billing.