Projects

Overview: Agent Deck is a terminal session manager designed for AI coding agents, allowing you to manage multiple sessions of different agents in a single terminal window. Key features: - Supports multiple agents like Claude, Gemini, OpenCode, and Codex. - Displays sessions in a single terminal window, showing running, waiting, or completed sessions. - Includes features like groups, search, forking, git worktrees, cost tracking, and a phone-controlled conductor. - Provides a seamless experience for managing a fleet of AI agents. Tech stack: Go Why it matters: Agent Deck simplifies the management of multiple AI agent sessions, making it easier to switch between projects and tasks, and keeping track of running sessions efficiently. Repository: asheshgoplani/agent-deck

### Overview: Make Money With AI is a curated list of open-source AI projects designed to help you generate income through the creation and monetization of AI-driven solutions. The repository includes projects that can be deployed across various operating systems and platforms, offering a wide range of tools and methodologies to automate workflows, build SaaS products, and offer enterprise integrations. ### Key Features: - **Openclaw**: Deploy a personal AI assistant across OS and platforms, automating workflows, selling SaaS agents, offering premium integrations, consulting, and enterprise support services. - **Rea**: Reverse engineer anything with agents, from app behavior down to native binaries. - **ECC**: Leverage battle-tested Claude configurations including agents, skills, hooks, commands, and MCPs to build revenue-generating AI products, automate workflows, accelerate deployment, enable enterprise integrations, reusable templates. - **Hermes-Agent**: Leverage autonomous agents, modular tool orchestration, persistent memory, and easy customization to automate services, build SaaS, and monetize AI-driven workflows for enterprise and consumer products. - **N8n**: Leverage the visual workflow builder, native AI, and 400+ integrations to automate processes, create SaaS products, sell automations, offer managed integrations, and upsell analytics services. - **Firecrawl**: Provides a scalable Web Data API to convert entire websites into LLM-ready markdown and structured data, enabling rapid productization, search, analytics, and recurring revenue monetization. - **AutoGPT**: Use autonomous agents, task automation, and

Overview: This repository contains demo programs for the "Talking Head Anime from a Single Image 2: More Expressive" project, allowing users to manipulate facial expressions and head rotations of an anime character from a single image using a graphical user interface and a Jupyter notebook. Key features: - Manual Poser: A graphical user interface for manipulating facial expressions and head rotations of an anime character. - IfacialMocap Puppeteer: A program that transfers facial motion captured by a commercial iOS application called iFacialMocap to an image of an anime character. Tech stack: Python 3, PyTorch, SciPY, wxPython, Matplotlib. Why it matters: The demo programs provide a practical way to explore and experiment with the Talking Head Anime technique, allowing users to create more expressive anime characters from single images.

Overview: VisionLLM Series is a series of large language models designed for vision-centric tasks, focusing on open-ended decoding and generalist multimodal capabilities. Key features: - VisionLLM: A large language model optimized for vision-centric tasks, capable of understanding and generating visual content. - VisionLLM v2: A generalist multimodal large language model designed to support a wide range of vision-language tasks, including visual understanding, perception, and generation. Tech stack: Python Why it matters: VisionLLM Series provides a powerful tool for researchers and developers looking to integrate vision-language capabilities into their applications, making it easier to tackle complex tasks that require both visual and textual understanding.

Overview: OC-SORT is a simple, online, and robust multi-object tracker designed for tracking in crowded scenes and non-linear motion, addressing limitations in Kalman filters and SORT. Key features: - Observation-centric re-update mechanism - Flexible integration with various detectors and matching modules - Simple, online, and real-time tracking Tech stack: Python Why it matters: OC-SORT improves tracking robustness in crowded scenes and non-linear motion, making it suitable for applications requiring high accuracy and real-time performance. Repository: noahcao/OC_SORT Primary language: Python GitHub stars: 1154

### Overview: This repository is a collection of evidence-backed use cases, prompts, integrations, evaluations, and safety notes for OpenAI's GPT-6 Astra model. It serves as a primary resource for developers looking to leverage the model's capabilities in practical applications. ### Key features: - Prompts and use cases - Integrations with various tools and platforms - Evaluations and performance metrics - Safety guidelines and best practices ### Tech stack: GitHub ### Why it matters: This repository provides a comprehensive overview of how GPT-6 Astra can be effectively used in real-world applications, ensuring developers have access to reliable and up-to-date information.

Overview: AI Agent Skills is a universal skill installer and package manager designed to streamline the development of AI coding agents. It simplifies the process of installing and managing a variety of skills, making it easier to create and maintain complex AI applications. Key features: - Bundled library of 12+ runtimes - Supports 18 skills from various skill sets - Provides a curated library of skills organized by shelves and provenance - Compatible with any Agent Skills-compatible agent Tech stack: JavaScript Why it matters: AI Agent Skills streamlines the installation and management of AI skills, making it easier for developers to create and maintain complex AI applications. The bundled library of 12+ runtimes and the curated skills provide a comprehensive solution for developers working with AI coding agents. Repository: MoizIbnYousaf/ai-agent-skills Primary language: JavaScript GitHub stars: 1152 GitHub description: Universal skill installer and package manager for AI coding agents. One command, 12+ runtimes. npx ai-agent-skills

### Overview: 本项目实现了YOLOv5与DeepSort的结合，用于目标检测和追踪。通过封装成Detector类，使得其更易于集成到其他项目中。项目支持训练自己的数据集，并提供详细的代码示例和运行指导。 ### Key features: - 使用YOLOv5进行目标检测 - 使用DeepSort进行目标追踪 - 支持5.0版本的YOLOv5模型训练 - 显示目标类别信息 - 支持5种目标类别：行人、汽车、卡车等 ### Tech stack: Python ### Why it matters: 该项目为开发者提供了一种高效、准确的目标检测和追踪解决方案，适用于各种应用场景，如自动驾驶、安防监控等。通过YOLOv5的高性能和DeepSort的精准追踪，实现了目标的实时识别和动态跟踪。

### Overview: fairseq2 is a sequence modeling toolkit designed for researchers to train custom models for content generation tasks. It allows for instruction finetuning and preference optimization, supports multi-GPU and multi-node training, and offers native support for vLLM. The toolkit is built on modern PyTorch tooling, including composability features like torch.compile and PyTorch FSDP. ### Key Features: - First-party recipes for instruction finetuning and preference optimization. - Multi-GPU, multi-node training using DDP, FSDP, and tensor parallelism, capable of handling models up to 70B parameters. - Native support for vLLM with built-in sampling and beam search sequence generators. - Extensible with setuptools extension mechanism, allowing researchers to register new models, optimizers, lr schedulers, and trainer units without forking the library. - Modern PyTorch tooling, including composability features. - Streaming-based, high throughput data pipeline API written in C++ with support for speech and video decoding. - Programmatic asset cards for version-controlled access to models, datasets, and tokenizers. - Flexible, but deterministic configuration based on the built-in structured API. ### Tech Stack: fairseq2 is built on modern PyTorch tooling and uses composability features like torch.compile and PyTorch FSDP. ### Why It Matters: fairseq2 is a complete new project rather than an incremental update to the original fairseq, reflecting its distinct and separate identity. It offers a clean, modular API that allows researchers to independently own their

### Overview: Plug and Play Language Model implementation, allowing users to steer the topic and attributes of GPT-2 models without requiring extensive training. ### Key features: - **PPLM (Plug and Play Language Model)**: A simple approach to controlled text generation, enabling researchers to leverage best-in-class LMs. - **PPLM-BoW (Plug and Play Language Model Bag-of-Words)**: Example command for bag-of-words control. - **PPLM-Discrim (Plug and Play Language Model Discriminator)**: Example command for discriminator-based sentiment control. - **Hyperparameter Tuning**: Options to increase `--stepsize` for intensifying topic control, adjust `--kl_scale` or `--gm_scale` for modifying control, and use `--grad-length` for controlling the model's length. ### Tech stack: Python ### Why it matters: Researchers can now easily steer the topic and attributes of GPT-2 models without the need for extensive training, making it accessible for a wide range of applications.

### Overview: Biniou is a self-hosted web interface for a variety of generative artificial intelligence models, allowing users to generate multimedia content and interact with a chatbot without requiring a dedicated GPU or high RAM. It supports multiple models and can be run offline once the necessary models are downloaded. ### Key Features: - **Model Support**: Supports a wide range of generative AI models including Chatbots, SDXL, Anime, MusicGen, and more. - **Self-Hosted**: Runs on GNU/Linux, OpenSUSE, RHEL, Arch, Mandriva, Debian, and Ubuntu. - **Offline Capabilities**: Works offline once the models are downloaded. - **Docker**: Available as a Docker image for easy deployment. - **Documentation**: Comprehensive documentation and a Showroom for visual demonstrations. ### Tech Stack: Python ### Why It Matters: Biniou simplifies the process of using generative AI without the need for expensive hardware, making it accessible to a broader audience. Its self-hosted nature allows for easy deployment and management, making it a valuable tool for both personal and professional use cases.

### Overview: This repository is a curated list of resources focused on Natural Language Processing (NLP) with a particular emphasis on Transformer networks, attention mechanisms, and large language models (LLMs). It includes papers, articles, tutorials, and more, providing a comprehensive guide for developers and researchers interested in the latest advancements in NLP. ### Key Features: - Papers: A collection of machine learning and deep learning resources for NLP. - Articles: Detailed articles on various NLP topics. - BERT and Transformer: A deep dive into Bidirectional Encoder Representations from Transformers. - Attention Mechanism: Exploring the core of NLP models. - Transformer Architecture: An overview of Transformer networks. - Generative Pre-Training Transformer (GPT): Focus on ChatGPT and its applications. - Large Language Model (LLM): Resources for understanding and implementing LLMs. - Transformer Reinforcement Learning: A look at how transformers are used in reinforcement learning. - Additional Reading: Extensive references for further exploration. - Educational: Tutorials and educational materials. - Tutorials: Comprehensive guides for beginners and advanced learners. - AI Safety: Resources for understanding and addressing potential risks in AI. - Videos: Educational videos on NLP topics. - BERTology: A series on the BERT model. - Attention and Transformer Networks: Exploring the attention mechanism and transformer architecture. - Official BERT Implementations: Resources for using the official BERT model. - Transformer Implementations: Community-driven implementations in various frameworks. - PyT

Overview: EvoTorch is an advanced evolutionary computation library built on PyTorch, designed for solving various optimization problems, including black-box optimization, reinforcement learning, and supervised learning tasks. Key features: - Supports both differentiable and non-differentiable problems - Provides a variety of evolutionary algorithms, including distribution-based and population-based methods - Integrates seamlessly with PyTorch for efficient computation and parallelization on GPUs - Utilizes Ray for distributed optimization across multiple CPUs, GPUs, and computers in a Ray cluster Tech stack: Python, PyTorch, Ray Why it matters: EvoTorch simplifies the process of implementing evolutionary algorithms, making it easier to solve complex optimization problems with PyTorch. Its ability to handle both differentiable and non-differentiable objectives makes it suitable for a wide range of applications, from machine learning to reinforcement learning. The integration with PyTorch and Ray's distributed capabilities ensures efficient performance, especially when leveraging GPUs and distributed computing resources.

### Overview: This repository aims to collect and consolidate resources related to Data-Centric AI, focusing on the perspective of making it easier for practitioners to understand, program, and iterate on datasets. It serves as a single, accessible place for anyone interested in this area to learn and contribute. ### Key Features: - Resources and progress in data-centric AI. - Exciting directions past, present, and future. - A journey to data-centric AI, including a blog post. - Why data-centric AI is important as a viewpoint for AI. - A class at Stanford on data-centric AI, seeking feedback. - Contributing guidelines for expanding the repository. ### Tech Stack: GitHub, LaTeX ### Why It Matters: This repository aims to bring together practical insights and experiences from those who have implemented data-centric AI in real-world scenarios, providing a comprehensive resource for anyone interested in this field. ### How Can I Help? - Join the mailing list for more information. - Fill out the short form to better understand interests. - Provide feedback on the class. - Submit ideas for improving the repository. ### Contributing - Welcome to contribute to this GitHub repository by reading the [contributing guidelines](CONTRIBUTING.md). ### Table of Contents - Background - Data Programming & Weak Supervision - Data Augmentation - Self-Supervision - The End of Modelitis - Fine-Grained Evaluation - Robustness - Data Cleaning - MLOps - Data Selection - Data Privacy (Under construction) - Data Flow (Under construction) - Multi-

### Overview: This project implements the "Painting Outside the Box" technique, focusing on image outpainting using Keras. The modifications include adding identity loss, removing patches, replacing masking with cropping, and incorporating convolution layers. ### Key Features: - Added identity loss to align generated images with the original. - Removed patches from the training data. - Replaced masking with cropping in the training pipeline. - Added convolution layers to the model. ### Tech Stack: Keras Implementation ### Why It Matters: This implementation showcases advanced techniques for image outpainting, demonstrating how to effectively integrate Keras for practical applications in computer vision and image processing.

Overview: This repository provides comprehensive roadmaps for various roles in the Data Science, Data Analytics, and Artificial Intelligence industries, based on the top job postings on LinkedIn and Indeed. Key features: - Data Analyst - Data Engineer - Data Scientist - AI Engineer - AI Agent Developer - Software Engineer with Machine Learning - Computer Vision Engineer - Generative AI Engineer - Machine Learning Engineer - NLP Engineer - Domain-Specific ML Topics for Researchers Tech stack: None specified Why it matters: These roadmaps are designed to help individuals understand the skills, tools, and knowledge required for each role, ensuring a focused and efficient learning process.

Overview: Hexabot v3 is an AI workflow automation platform that integrates workflows, actions, agents, and conversational channels into a single runtime environment. Key features: - YAML workflow definitions with typed runtime contracts - Action-based execution with schema-validated inputs/outputs/settings - Reusable capability/config bindings separated from task logic - Explicit memory definitions and runtime memory integration - Model Context Protocol support for tool/context interoperability - Channels and helpers remain core concepts - Schema-first architecture for broad validation and shared contracts Tech stack: TypeScript Why it matters: Hexabot v3 combines AI capabilities with traditional workflow automation, making it a powerful tool for automating processes across multiple channels and platforms.

Overview: This project involves the development of a vehicle detection system using machine learning and computer vision techniques. The system utilizes Linear SVM, HOG feature extraction, color space conversion, and other methods to accurately identify vehicles in images. Key features: - Vehicle detection using machine learning and computer vision techniques. - Implementation of Linear SVM. - HOG (Histogram of Oriented Gradients) feature extraction. - Color space conversion and space binning. - Histogram of color extraction. - Sliding Window method for object detection. Tech stack: Python 3.4, Jupyter Notebook, LinearSVC, StandardScaler, Sklearn, OpenCV, Numpy, Matplotlib, Scikit-learn. Why it matters: This project is crucial for Udacity's Self-Driving Car Engineer Nanodegree, providing a practical application of machine learning and computer vision techniques in real-world autonomous vehicle development.

Overview: AI Factory simplifies AI integration and development, allowing developers to focus on code without worrying about setup and context. Key features: - Zero configuration setup - Best practices built-in for logging, commits, and code review - Spec-driven development with AI following plans - Community skills ecosystem - Stack-agnostic support - Multi-agent support with various AI agents Tech stack: Shell Why it matters: AI Factory streamlines the process of integrating AI into development workflows, enabling developers to focus on building quality code without the hassle of configuration and setup. Repository: lee-to/ai-factory GitHub stars: 1142

Overview: Lingua is a Rust library designed for accurate natural language detection, suitable for short text and mixed-language text. It simplifies the preprocessing step for applications like text classification and spell checking. Key features: - Supports 75 languages - Detects language in short text snippets - Combines rule-based and statistical Naive Bayes methods - Does not require external APIs or neural networks Tech stack: Rust Why it matters: Lingua provides accurate results on both short and long texts, eliminating the need for lengthy text fragments and improving detection accuracy even for single words and phrases. It is designed to be used offline, making it suitable for applications that require quick and reliable language detection without the need for machine learning frameworks or extensive configuration.