Trending AI projects

OmniChar Studio is an open‑source Python application that lets you build a portable `.char` file describing a character’s face, body, clothing and voice. The same character can be used with many generative models (FLUX, Krea, MiniMax H3, LTX‑2.5, etc.) for image or video generation, with optional LoRA training. It provides a web UI, a CLI installer that auto‑selects the correct CUDA/PyTorch wheel, low‑VRAM and CPU profiles, and a ComfyUI node for workflow integration. Licensed under GPL‑v3 and supported on NVIDIA, Apple Silicon, AMD ROCm, and CPU‑only setups.

A collection of 11 autonomous SEO skills for Claude and other AI agents that perform live research, competitor analysis, and content auditing without requiring manual data exports.

UniLab is an open‑source Python framework that lets you define robot reinforcement‑learning tasks once (via Hydra YAML) and run them on many physics simulators (MuJoCo, Isaac Gym, Motrix, etc.) and hardware backends (CPU, CUDA, ROCm, Apple Silicon, Intel XPU). It provides a `SimBackend` contract, off‑policy replay‑accelerated training (SAC/Flash‑SAC), and a support matrix for verified combinations. Install with `uv`, run demos, train/eval with simple CLI flags, and extend via documented contracts. Licensed Apache‑2.0 and cited in a CoRL 2026 paper.

Mission Planner is a ground control station software for ArduPilot-based unmanned vehicles, providing tools for mission planning, vehicle configuration, and firmware management.

A PowerPoint plugin for researchers that automates scientific image processing, layout alignment, and the insertion of LaTeX and Markdown content.

A library of 35 modular markdown-based skills and hooks for AI coding agents to standardize development workflows and reduce prompt bloat.

An open-source, node-based canvas that lets users chain text, image, and video AI models from 11 providers into automated, visual creation workflows.

A webcam-based hand-tracking interface that lets you manipulate notes, images, and 3D models spatially, designed to be wired into an AI agent as a visual and interactive body.

Versus Incident is an open‑source, self‑hosted SRE assistant that uses an AI‑style log‑pattern learner to auto‑detect novel errors and can also receive alerts via webhook. It provides a web dashboard with chat‑style incident summaries, Kubernetes cluster insights, multi‑channel notifications, and on‑call integrations (PagerDuty, AWS Incident Manager). Configuration is YAML‑based, the agent runs in training/shadow/detect modes, and the whole system is packaged as a Docker image (with optional Helm chart).

claude-history is a terminal‑based UI for searching, viewing, and managing Claude Code, Pi and Oh My Pi chat transcripts. It offers fuzzy and ONNX‑powered semantic search, in‑viewer markdown rendering, and one‑key actions to resume, fork, rename or delete sessions, making AI‑agent conversation history a searchable, actionable part of a developer’s workflow.

OKX Agent Trade Kit is an open‑source, AI‑enabled toolkit that lets LLM assistants (Claude, Cursor, etc.) control an OKX crypto account via the Model Context Protocol. It provides a local MCP server (`okx-trade-mcp`) exposing 170+ trading tools across market data, spot/swap/futures/options, bots, earn products, news, and smart‑money analytics, plus a matching CLI (`okx-trade-cli`). The system runs locally, keeping API keys private, and includes safety flags, ready‑made AI‑agent skill files, and MIT licensing.

An open-source AI video workstation that combines a full timeline editor with AI generation workflows via ComfyUI and MCP-based agent automation.

An AI agent skill for legal contract review that provides risk detection, market benchmarks, and specific redline suggestions grounded in the CUAD dataset.

A tool for running multiple AI coding agents in parallel using isolated, sandboxed virtual environments hosted locally or in the cloud.

A universal multi-engine TTS and audio suite for ComfyUI that integrates voice synthesis, voice conversion, and SRT subtitle synchronization across dozens of AI models.

This is a curated, continuously evolving hub that collects papers, benchmarks, blogs, podcasts, and other resources on self-improving agentic systems. It organizes the field into two main branches—improving the underlying model and improving the surrounding scaffolding—and serves as a practical guide for researchers and practitioners.

A curated collection of papers, frameworks, and tools for on-policy distillation, a post-training technique that trains student LLMs on their own generated samples to bridge the train-inference distribution gap.

A structured, community-maintained collection of papers, tools, and concepts about giving LLMs lasting memory—includes 1,270+ papers, 111 open-source projects, a screening policy for what counts, and regular updates.

A native AI keying runtime and OFX plugin for DaVinci Resolve and Foundry Nuke that provides automated, ML-accelerated background removal for video editors.

React Native ExecuTorch is a library that brings Meta’s on‑device ExecuTorch inference runtime to React Native, enabling offline AI (LLMs, vision, speech, etc.) with hardware acceleration and ready‑made hooks or low‑level TypeScript pipelines.

Google DeepMind’s TAP‑Net repository offers datasets, pretrained models (TAPIR, BootsTAPIR, TAPNext/Next++), training code, and Colab/real‑time demos for the “Tracking Any Point” task—precise point‑level video tracking useful for vision research and robot imitation.

An open-source Python library for researching communication systems, featuring tools for radio propagation ray tracing and wireless/optical link-level and system-level simulation.

SAHI is a Python library that adds tiled (sliced) inference to existing object‑detection models, enabling accurate detection of small objects in very large images or videos. It supports many popular detectors, offers a CLI and FiftyOne integration, and includes COCO‑dataset utilities.

DeepSpeed is a PyTorch‑based open‑source library that provides memory‑efficient optimizers, multi‑dimensional parallelism, and hardware‑offload engines to train extremely large deep‑learning models (up to hundreds of billions of parameters) on GPUs, CPUs, and specialized accelerators.

An open-source MLOps and LLMOps suite that provides experiment tracking, orchestration, and data management to streamline the deep learning development lifecycle.

An AI-powered tool that automatically generates unit tests to increase code coverage, supporting multiple programming languages and various LLM providers.

OpenLLM is a Python tool that lets you run open‑source LLMs (e.g., Llama 3, Qwen 2.5, Mistral) locally and expose them through OpenAI‑compatible REST APIs. With a single command you can start a GPU‑accelerated server, use a built‑in web chat UI, and deploy the service via Docker or Kubernetes. It supports a catalog of models, custom model repositories, and integrates with the high‑performance vLLM backend.

SkillRL is a framework that enables LLM agents to learn and evolve a hierarchical library of reusable behavioral skills from past experiences, reducing token overhead and improving RL performance.

An agent-native toolkit for Kalshi prediction markets that focuses on rigorously measuring and proving trading edge through settlement-grounded telemetry and a self-improving risk policy.

A comprehensive simulation benchmark for testing embodied AI agents on 1,000 everyday household activities based on real human time-use surveys.

A PyTorch-native toolkit for reinforcement learning and robotics that uses a composable, TensorDict-based data model to build scalable RL systems.

LLM Vision is a Home Assistant integration that uses multimodal large‑language models (OpenAI, Anthropic, Gemini, Ollama, etc.) to analyze images, video clips, live camera feeds and Frigate events. It returns natural‑language descriptions, remembers identified people/pets/objects, and stores each analysis in a timeline that can be shown on the dashboard or queried via Assist. Install via HACS, configure an LLM provider, and optionally add the supplied Blueprint for AI‑summarized camera notifications.

A weekly live-coding series that teaches developers how to move AI applications from initial prototypes to reliable, production-ready systems.

A comprehensive educational repository providing structured tutorials, Jupyter Notebooks, and exercises for learning Python, data science, and machine learning from beginner to advanced levels.

A 17-phase, project-driven syllabus designed to take a beginner to a production-grade AI engineer, covering Python, math, ML/DL, LLMs, orchestration, RAG, agents, MLOps, fine-tuning, and ethics, with 51 hands-on projects and role-specific starting points.

The repository contains a computer-vision book written in Quarto Markdown. To work on it, you install Quarto, run `quarto preview`, and edit `.qmd` files. Because the repo is big, a shallow clone is recommended.

A multi-agent AI writing assistant that turns article creation into an interactive, reviewable process: a writing director, researcher, main writer, editors, reader simulators, and fact-checkers collaborate with you stage by stage. Ships as a Windows desktop app (RC) and as a Claude Code Skill, with 99 model presets, optional web search for fact-checking, local versioned storage, and full run logs.

A high-performance library of deep learning-based denoising filters that reduces rendering times for ray tracing applications by removing Monte Carlo noise.

Writing Tools is a free, open‑source desktop app for Windows, Linux and macOS that lets you invoke an LLM anywhere on your computer with a hot‑key. It can proofread, rewrite, change tone, translate, summarise web pages or YouTube transcripts, and even chat with the model. You can use cloud APIs (Gemini, OpenAI, Anthropic, etc.) or run local models via Ollama, llama.cpp or Apple‑silicon MLX. The app is privacy‑focused (no telemetry, local storage of API keys) and licensed under GPL‑v3.

A structured AI-assisted development framework for Claude Code that uses a Plan-Apply-Unify loop to prevent context rot and ensure high-quality, verified execution.

GeoCalib is a Python library that estimates camera intrinsics and gravity direction from a single image using deep learning plus geometric optimisation. It supports several lens models, works with partial priors, can calibrate batches or multi‑camera rigs, and comes with training/evaluation code (OpenPano dataset) and demos. It outperforms prior single‑image calibration methods on standard benchmarks.

A persistent memory layer for AI agents featuring Git-like version control, allowing for snapshots, branching, and hybrid semantic search to maintain auditable long-term knowledge.

An enterprise-grade AI agent platform that implements a complete loop from document governance and hybrid RAG retrieval to ReAct-based agentic reasoning and MCP tool integration.

A self-hosted agentic framework for creating customizable AI personas with voice interaction, long-term memory, and autonomous system control.

A comprehensive setup for Claude Code that integrates Obsidian and Graphify to provide persistent long-term memory and token-efficient codebase awareness.

T-Rex is a tactile-reactive dexterous manipulation framework that uses an asynchronous Mixture-of-Transformers to react dynamically to high-frequency touch signals for delicate robotic tasks.

An academic harness for AI agents that provides the structured evidence, library management, and repeatable workflows needed for professional research and academic writing.

An Unreal Engine 5-based simulation platform for developing and evaluating LLM/VLM agents in realistic physical and social environments.