Projects

### Overview: Agents Towards Production is a comprehensive resource for developers looking to build production-grade GenAI agents. This project provides tutorials on stateful workflows, vector memory, real-time web search APIs, Docker deployment, FastAPI endpoints, security guardrails, GPU scaling, multi-agent coordination, observability, evaluation, and UI development. ### Key Features: - **Stateful Workflows**: Tutorials on how to build and manage stateful AI agents. - **Vector Memory**: Techniques for using vector memory in AI applications. - **Real-Time Web Search APIs**: How to integrate real-time web search APIs into your AI agent. - **Docker Deployment**: Best practices for deploying AI agents using Docker. - **FastAPI Endpoints**: Tutorials on creating and managing FastAPI endpoints. - **Security Guardrails**: Best practices for securing AI agents. - **GPU Scaling**: How to scale AI agents using GPUs. - **Multi-Agent Coordination**: Techniques for coordinating multiple AI agents. - **Observability**: How to monitor and debug AI agents. - **Evaluation**: Metrics and methods for evaluating AI agents. - **UI Development**: Tutorials on building user interfaces for AI agents. ### Tech Stack: - **Jupyter Notebook**: The primary tool used for developing and documenting tutorials. ### Why It Matters: This project is essential for developers looking to build production-grade AI agents. It provides a structured approach to turning AI prototypes into enterprise-level products, covering everything from the foundational concepts to advanced deployment strategies. By following these tutorials, developers can ensure that their AI agents are reliable, efficient, and scalable

Overview: Coze Studio is an all-in-one AI agent development platform that simplifies the creation, debugging, and deployment of AI agents, making it easier and more accessible than ever before. Key features: - Model service: Manages model integration with services like OpenAI and Volcengine. - Build agent: Supports building, publishing, and managing agents, including workflows, knowledge bases, and other resources. - Build apps: Enables the creation and publishing of apps, and the development of business logic through workflows. - Build a workflow: Allows the creation, modification, publishing, and deletion of workflows. - Develop resources: Supports creating and managing plugins, knowledge bases, databases, and prompts. - API and SDK: Provides tools for creating conversations, initiating chats, and integrating agents or apps into your own app through Chat SDK. Tech stack: Coze Studio is built using Golang for the backend and React with TypeScript for the frontend, following microservices and domain-driven design principles. Why it matters: Coze Studio simplifies AI agent development, enabling developers to focus on creating the core value of AI without worrying about the complexities of backend infrastructure. It offers a low-code approach to AI app development, making it an ideal choice for building low-code AI products tailored to specific business needs.

Overview: Ciphey is a powerful tool that automatically decrypts encryptions, decodes encodings, and cracks hashes without requiring a key or cipher name. It supports a wide range of encoding methods, including Base64, hex, Caesar/ROT13, Vigenère, and more. Key features: - Decrypts Base64, hex, Caesar/ROT13, Vigenère, and several other encoding methods. - No need for a key or cipher name. - Supports multiple layers of encoding. - Fast and efficient. Tech stack: Rust. Why it matters: Ciphey is essential for anyone dealing with encrypted or encoded text, providing a convenient and powerful solution for decoding and cracking hashes without the need for manual intervention. Repository: bee-san/Ciphey GitHub stars: 21684 GitHub description: ⚡ Automatically decrypt encryptions without knowing the key or cipher, decode encodings, and crack hashes ⚡

Overview: HivisionIDPhotos is a lightweight and efficient AI ID photos tools designed for producing standard and customized passport-sized photos. Key features: - Supports both CPU and GPU inference for quick and accurate results. - Offers multiple photo sizes for various printing needs. - Includes a variety of features such as face alignment, printing layout options, and more. - Comes with a Gradio demo for easy experimentation and a Docker deployment option for production use. Tech stack: Python, TensorFlow, PyTorch, and other AI libraries. Why it matters: This tool simplifies the process of creating professional-looking ID photos, making it easier for users to obtain valid and professional-looking documents without requiring complex hardware or extensive training. Repository: Zeyi-Lin/HivisionIDPhotos Primary language: Python GitHub stars: 21,642 GitHub description: ⚡️HivisionIDPhotos: a lightweight and efficient AI ID photos tools. 一个轻量级的AI证件照制作算法。 README excerpt: <div align="center"> <img alt="hivision_logo" src="assets/hivision_logo.png" width=120 height=120> <h1>HivisionIDPhoto</h1> [English](README_EN.md) / 中文 / [日本語](README_JP.md) / [한국어](README_KO.md) [![][release-shield]][release-link] [![][dockerhub-shield]][dockerhub-link] [![][github-stars-shield]][github-stars-link] [

Overview: ONNX, Open Neural Network Exchange, is an open standard format for machine learning models, supporting both deep learning and traditional ML. It provides an extensible computation graph model and defines built-in operators and data types, enabling interoperability between different frameworks and tools. Key features: - ONNX models can be created using Python, making it accessible to developers. - ONNX models can be easily converted between different frameworks, facilitating the transition from research to production. - ONNX supports a wide range of tools and frameworks, including popular ones like TensorFlow, PyTorch, and Hugging Face. Tech stack: Python Why it matters: ONNX simplifies the process of model interoperability and enables developers to choose the right tools as their projects evolve, accelerating innovation in the AI community. Repository: onnx/onnx GitHub stars: 21594 GitHub description: Open standard for machine learning interoperability

Overview: 数字生命卡兹克开源的AI技能合集，包括领导、存储分析器、AI HOT资讯查询、洁癖、横纵分析法等实用功能。这些技能遵循开放标准，适用于支持Agent Skills的工具。 Key features: - Leader（领导）：将模糊的想法定义为清晰的目标，让AI独立完成任务。 - Storage Analyzer（存储分析器）：扫描Mac/Windows整机磁盘，三色分级清理决策。 - AI Hot（AI HOT）：提供每日AI HOT动态和AI资讯查询。 - Neat-Freak（洁癖）：自动对齐项目文档、CLAUDE.md和Agent记忆。 - HV Analysis（横纵分析法）：研究产品/公司/概念的万字研究报告。 - Khazix Writer（卡兹克写作）：使用我的口吻和节奏写公众号长文。 Tech stack: Python Why it matters: 这些技能是基于开放标准的AI指令集，旨在提高AI的自定义能力和协作效率，特别是在复杂任务和长程目标上。

### Overview: FinGPT is an open-source financial large language model project developed by the AI4Finance Foundation, aiming to democratize internet-scale data for financial large language models. ### Key Features: - FinGPT-Forecaster: A financial forecaster model released on HuggingFace. - Financial Multi-Task LLMs: Financial large language models produced for evaluating base-LLMs on FinGPT-Benchmark. - Financial Sentiment Analysis Model: A model for enhancing financial sentiment analysis using retrieval augmented large language models. ### Tech Stack: Jupyter Notebook ### Why it Matters: FinGPT is crucial for democratizing internet-scale data for financial large language models, making it easier for researchers and practitioners to access and utilize financial data in their applications.

### Overview: Machine Learning for Trading, 3rd Edition is a comprehensive guide that covers the entire process of building, testing, and deploying machine learning-driven trading strategies from data sourcing to live execution. ### Key Features: - **End-to-End Workflow**: Traces one process from data infrastructure and strategy research to deployment and monitoring, including a feedback loop for retraining, pausing, or retiring strategies. - **Model Toolkit**: Includes gradient boosting, deep time-series architectures, and newer tabular and latent-factor models. - **Strategy Design**: Full chapters on transaction costs, risk management, portfolio construction, and strategy synthesis. - **Production Track**: Covers live trading systems, MLOps, governance, and operational reality. - **Generative AI**: Uses retrieval-augmented generation and autonomous research systems. - **Causal ML**: Double ML, Bayesian structural time series, and causal discovery. - **Reinforcement Learning**: Optimal execution, market making, and deep hedging. - **Synthetic Data**: TimeGAN, Tail-GAN, Sig-CWGAN, and diffusion-based generators. ### Tech Stack: Jupyter Notebook ### Why It Matters: The book provides a structured approach to building, testing, and deploying machine learning-driven trading strategies, ensuring methodological rigor and practical application in real-world scenarios.

Overview: Rasa is an open-source machine learning framework designed for automating text- and voice-based conversations. It supports integration with various platforms like Facebook Messenger, Slack, and more. Key features: - Supports integration with multiple platforms including Facebook Messenger, Slack, and Google Hangouts. - Enables building of contextual assistants that can handle layered conversations. - Provides a conversational AI engine called CALM, combining language models with strict business logic. Tech stack: Python Why it matters: Rasa simplifies the process of building conversational AI applications, making it accessible for developers without extensive machine learning expertise.

### Overview: Agency Agents 中文版是一款集成了277个AI专家角色的平台，支持Claude Code、Cursor、Copilot等20种工具，覆盖工程、设计、营销、金融等20个部门。搭配编排器agency-orchestrator，一句话即可让多位专家按DAG自动协作。 ### Key features: - 277个即插即用的AI专家角色 - 支持20种AI编程工具 - 20个部门覆盖 - 64个中国市场原创智能体 ### Tech stack: Shell ### Why it matters: 通过这个平台，用户可以更高效地管理和协作多个AI专家，从而提升工作效率和质量。

### Overview: 鱼皮的AI知识库是一个完全免费开放的AI知识共享平台，汇集了当前热门的AI工具相关信息，包括产品介绍、使用指南、工具测评、技巧分享、应用场景、AI变现、行业资讯、教程资源等一系列内容。旨在帮助任何人快速掌握AI技术，走在时代前沿。 ### Key features: - AI工具介绍 - 使用指南 - 工具测评 - 技巧分享 - 应用场景 - AI变现 - 行业资讯 - 教程资源 - 项目实战 - 产品开发 - 技术选型 - 架构设计 - 盈利模式 - SEO优化 - 自媒体运营 ### Tech stack: JavaScript - 项目语言: JavaScript - 项目类型: 开源文档 ### Why it matters: 鱼皮的AI知识库通过整合各种AI工具和资源，帮助程序员、设计师、产品运营等不同背景的用户快速掌握AI技术，打破信息壁垒，让每个人都能平等获取AI时代的工具与认知，利用科技让生活更美好。

Overview: This project provides a text-to-CAD plugin that enables agents to generate 3D models in various formats such as STEP, GLB, STL, and 3MF. It also offers design for manufacturing checks, engineering drawings, and integration with popular 3D printing, sheet metal, and CNC fabrication services. Key features: - Text-to-CAD plugin for generating 3D models - Design for manufacturing checks - Engineering drawings - Integration with 3D printing, sheet metal, and CNC fabrication services - Supported by popular agents like Claude Code, Codex, Cursor, Gemini, and Grok Tech stack: Python Why it matters: This plugin allows agents to leverage CAD capabilities directly within their workflows, enhancing their design and manufacturing processes. It streamlines the creation of 3D models and provides essential design checks, making it a valuable tool for professionals in the field. Repository: earthtojake/text-to-cad

### Overview: This repository is a curated list of awesome open source libraries designed to help you deploy, monitor, version, scale, and secure your production machine learning projects. It aims to provide a comprehensive toolkit for machine learning operations, covering various aspects such as AutoML, data pipelines, data stream processing, model training, and more. ### Key features: - AutoML (Automated Machine Learning) - Data Pipeline - Data Science Notebook - Data Storage Optimization - Data Stream Processing - Deployment & Serving - Explainability & Fairness - Feature Store - Industry-strength Anomaly Detection - Industry-strength Computer Vision - Industry-strength Information Retrieval - Industry-strength Natural Language Processing - Industry-strength Recommender System - Industry-strength Reinforcement Learning - Industry-strength Robotics - Industry-strength Visualisation - Metadata Management - Model, Data & Experiment Management - Model Storage Optimization - Model Training & Orchestration - Privacy & Safety ### Tech stack: The repository is built using Markdown, HTML, and various open-source libraries and tools. ### Why it matters: This repository is essential for anyone involved in machine learning operations, providing a one-stop-shop for all the tools needed to deploy, monitor, and scale production machine learning projects. It helps ensure that your models are secure, efficient, and transparent, making your machine learning projects more effective and reliable.

### Overview: FunASR is an open-source speech recognition toolkit designed for training, inference, and deployment of speech recognition models. It includes features such as ASR, VAD, punctuation, speaker diarization, and OpenAI-compatible serving. ### Key features: - **ASR (Automatic Speech Recognition)**: Transcribes audio files into text. - **VAD (Voice Activity Detection)**: Identifies and segments speech segments in audio. - **Punctuation**: Adds punctuation marks to transcribed text. - **Speaker Diarization**: Separates speech into different speakers. - **Emotion and Audio-Event Models**: Analyzes emotions and audio events. - **OpenAI-compatible Serving**: Provides a model serving endpoint compatible with OpenAI's format. ### Tech stack: Python ### Why it matters: FunASR is essential for developers and researchers looking to integrate speech recognition capabilities into their projects. It simplifies the process of training, deploying, and using speech recognition models, making it a valuable tool for applications in various domains such as customer service, healthcare, and language translation.

### Overview: Hummingbot is an open-source framework designed to help developers create and deploy automated trading strategies, or bots, that can run on various exchanges. ### Key Features: - **Paper Trading Script**: Simulates trading against live market data without requiring any API keys. - **CLI Interface (`hbot`)**: A command-line interface for running and controlling bots non-interactively. - **Condor (AI Harness)**: A tool for building and running automated strategies that leverage AI decision-making. - **GitHub Stars**: 20,526 stars, indicating widespread adoption and interest in the project. ### Tech Stack: Python ### Why it Matters: Hummingbot democratizes high-frequency trading by providing a community-driven framework for algorithmic traders and developers, enabling the creation and deployment of automated trading strategies across multiple exchanges.

### Overview: SkillSpector is a security scanner designed to identify vulnerabilities, malicious patterns, and security risks in AI agent skills before installation. It helps developers ensure the safety of skills before deployment. ### Key Features: - **Multi-format Input**: Supports scanning of Git repos, URLs, zip files, directories, and single files. - **71 Vulnerability Patterns**: Covers 17 categories including prompt injection, data exfiltration, privilege escalation, supply chain, excessive agency, output handling, system prompt leakage, memory poisoning, tool misuse, rogue agent, anti-refusal, trigger abuse, dangerous code (AST), taint tracking, YARA signatures, MCP least privilege, and MCP tool poisoning. - **Two-Stage Analysis**: Fast static analysis followed by optional LLM semantic evaluation. - **Live Vulnerability Lookups**: Supports SC4 queries for real-time CVE data with offline fallback. - **Multiple Output Formats**: Reports in terminal, JSON, Markdown, and SARIF formats. - **Risk Scoring**: Scores between 0-100 with severity labels and clear recommendations. - **Baseline / False-Positive Suppression**: Accepts known findings via glob rules or fingerprint baselines to surface only new issues. ### Tech Stack: Python ### Why It Matters: SkillSpector helps developers ensure the safety of AI agent skills before installation, reducing the risk of vulnerabilities, malicious patterns, and security risks that could compromise the system. It is part of the NVIDIA Verified Skills pipeline, ensuring skills are vetted and signed before publication.

### Overview: This project-based course provides a comprehensive introduction to building the foundational components of AI coding agents, including environment setup, state management, verification, and control mechanisms. It covers 15 languages, making it accessible to developers worldwide. ### Key Features: - **Frontier Harness Design Breakdowns**: Apply the course's five-subsystem framework to reverse-engineer how four leading AI products build real harnesses. - **8 Projects**: Hands-on projects to apply the concepts learned in real-world scenarios. - **15 Languages**: The course is available in 15 languages, including English, 简体中文, 繁體中文, 日本語, 한국어, Español, Français, Русский, Deutsch, العربية, Tiếng Việt, Oʻzbekcha, Türkçe, and Portuguese (BR). ### Tech Stack: TypeScript ### Why It Matters: This course is essential for developers looking to build reliable and robust AI coding agents. It covers the essential components of AI development, making it a valuable resource for anyone interested in the field of artificial intelligence.

### Overview: This repository compiles essential concepts from Stanford's CS 229 Machine Learning course, including refreshers on prerequisites, cheatsheets for various machine learning fields, and a comprehensive compilation of tips and tricks. It aims to provide a one-stop resource for all learners. ### Key Features: - **VIP Cheatsheets**: Comprehensive guides covering supervised learning, unsupervised learning, deep learning, and machine learning tips and tricks. - **VIP Refreshers**: Quick review sections on probabilities and statistics, algebra and calculus. - **Super VIP Cheatsheet**: A single, all-inclusive compilation of the above elements. ### Tech Stack: Unknown ### Why It Matters: This repository serves as a valuable reference for students and professionals in machine learning, providing a consolidated resource to enhance understanding and application of the course material.

Overview: Kortix is an open-source AI Operating System designed to replace Claude Cowork and ChatGPT Work for teams, providing an alternative to closed-source alternatives. It offers agents that work on real cloud computers, with versioning, diffability, and shared management by the company. Key features: - Agents that deliver finished work - AI agents in Slack or Microsoft Teams - Background coding agents in the cloud - Replace Zapier-style automations with AI agents - AI Operating System for a company with agents - Governed agents with per-tool rules, approval gates, and an audit trail Tech stack: TypeScript Why it matters: Kortix provides an open-source alternative to closed-source AI alternatives, allowing teams to control their own AI agents and infrastructure. Repository: kortix-ai/suna Primary language: TypeScript GitHub stars: 20280 GitHub description: The open-source AI Operating System

Overview: DB-GPT is an open-source AI data assistant designed to connect to various data sources, write SQL autonomously, and execute Python and code-driven analysis workflows. It also provides tools for generating reports, insights, and action plans. Key features: - Connects to databases, CSV/Excel files, warehouses, and knowledge bases. - Writes SQL autonomously when asked natural language questions. - Runs Python- and code-driven analysis workflows. - Loads and executes reusable skills for domain-specific tasks. - Generates charts, dashboards, HTML reports, and analysis summaries. - Executes tasks safely in sandboxed environments. Tech stack: Python Why it matters: DB-GPT is a comprehensive solution for AI and data products, providing a powerful platform for building AI-native data agents and workflows, as well as a robust toolset for generating actionable insights and reports. Repository: eosphoros-ai/DB-GPT Primary language: Python GitHub stars: 20,158 GitHub description: open-source agentic AI data assistant for the next generation of AI + Data products.