Coding & Development

STUD is an open-source AI coding assistant specifically designed for Roblox Studio, offering deep integration to streamline development workflows. It enables developers to write and edit Luau scripts, manipulate game instances, and query DataStores directly from their terminal interface. Beyond Roblox-specific tasks, STUD functions as a general coding assistant, supporting file read/write/edit, glob and grep searches, and bash execution. The tool features a robust permission system, ensuring users approve sensitive actions like writes and shell commands, maintaining full control over every session. STUD is model-flexible, allowing users to configure different model providers and settings to suit their needs and budget, making it a versatile and powerful solution for Roblox developers.

SoftSpell, formerly CodeSpell, is an AI-powered platform designed to accelerate and streamline the entire Software Development Life Cycle (SDLC) for enterprises. It offers a suite of tools including ReqSpell for requirements extraction and tracing, CodeSpell for AI-assisted code generation and documentation, and TestSpell for automated test case creation and validation. The platform focuses on modernizing legacy systems by transforming outdated applications, fragmented documentation, and rigid testing into scalable, AI-accelerated solutions. Key benefits include faster time-to-market, improved code consistency, reduced modernization risks, and predictable development timelines. SoftSpell integrates seamlessly across various IDEs, languages, and deployment pipelines, making it a comprehensive co-pilot for engineering teams.

php-nlp-tools is an open-source collection of Natural Language Processing (NLP) tools specifically designed for PHP 5.3+ environments. It enables developers to integrate advanced text analysis capabilities into their PHP applications. The library includes classification models like Multinomial Naive Bayes and Maximum Entropy, as well as experimental Topic Modeling with Latent Dirichlet Allocation. For text processing, it offers various tokenizers such as WhitespaceTokenizer and PennTreebankTokenizer, alongside stemmers like PorterStemmer and GreekStemmer. Additionally, it provides utilities for similarity calculations (Jaccard Index, Cosine similarity) and optimizers for MaxEnt models, including a fast, parallel gradient descent optimizer written in Go. This comprehensive toolkit is ideal for developers looking to implement NLP features directly within their PHP projects.

how-to-optim-algorithm-in-cuda is a comprehensive open-source repository dedicated to optimizing algorithms using CUDA. It offers a wealth of resources including code implementations for fundamental CUDA operators like reduce, softmax, and elementwise operations, as well as detailed learning notes and blog translations related to GPU and large language models. The project covers advanced topics such as CUTLASS, CuTe DSL, Triton, and PTX ISA, making it an invaluable learning tool for developers aiming to enhance the performance of their CUDA code. It also includes notes on large language model inference/training optimization and GPU/AI system papers.

kosong serves as an LLM abstraction layer specifically designed for modern AI agent applications, facilitating the seamless integration of large language models. This open-source project, hosted on GitHub, provides developers with a foundational toolkit to build sophisticated AI agents. The ongoing development of kosong has been migrated to the kimi-cli monorepo, indicating its continued evolution and integration within a broader ecosystem. It aims to streamline the process of leveraging LLMs, making it easier for developers to incorporate advanced AI capabilities into their applications and agents.

Patched offers reliable AI solutions for regulated operations, focusing on execution rather than just orchestration. It enables businesses to automate complex workflows in areas such as fraud and risk management, banking customer operations, and reconciliation. The platform is designed for full auditability and control, with systems deployed on dedicated infrastructure to ensure compliance. Patched takes responsibility for outcomes, not just system functionality, and has demonstrated significant ROI and resolution rates in various use cases. It is trusted by publicly listed enterprises to handle high-sensitivity workflows, drastically reducing turnaround times and improving SLA compliance.

PIXIU is an open-source resource designed to support the development, fine-tuning, and evaluation of Large Language Models (LLMs) specifically for the financial domain. It introduces the first financial LLMs, instruction tuning data, and evaluation benchmarks to holistically assess these models. The project is structured into key components: FinBen, a financial language understanding and prediction evaluation benchmark; FIT, a multi-task and multi-modal instruction dataset tailored for financial tasks; and FinMA, the core financial LLM. PIXIU offers open resources, multi-task and multi-modality capabilities, and diverse evaluation benchmarks, including critical financial prediction tasks aligned with real-world scenarios, making it a comprehensive platform for financial AI research and development.

Lasagne is a lightweight, open-source library designed to simplify the process of building and training neural networks using Theano. It supports a wide range of network architectures, including feed-forward networks like Convolutional Neural Networks (CNNs) and recurrent networks such as Long Short-Term Memory (LSTM), allowing for complex combinations. The library facilitates architectures with multiple inputs and outputs, including auxiliary classifiers. It integrates numerous optimization methods like Nesterov momentum, RMSprop, and ADAM, and allows for freely definable cost functions, leveraging Theano's symbolic differentiation to avoid manual gradient derivation. Lasagne also provides transparent support for both CPUs and GPUs through Theano's expression compiler, making it a versatile tool for deep learning projects.

NTU-Machine-learning is an open-source repository providing comprehensive materials for a machine learning course taught by Professor Hung-yi Lee at National Taiwan University. It includes lecture slides, detailed notes, and video links covering core topics such as supervised learning (regression, classification), unsupervised learning (AutoEncoder, Deep Generative Model), transfer learning, and structured learning. The repository also features weekly assignments with code examples, environment configuration guides for Anaconda and Docker, and recommendations for essential resources like blogs, classic deep learning papers, and Python tutorials. It's designed to offer a structured, self-paced learning experience for individuals looking to master machine learning concepts and practical applications.

MobileNet-CoreML provides an implementation of the MobileNet neural network architecture, as detailed in the paper "MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications," specifically tailored for Apple's CoreML framework. This tool facilitates the integration of efficient neural networks into iOS applications, leveraging pretrained weights from shicai/MobileNet-Caffe. It includes two demo apps: a Cat Demo for static image prediction and a Camera Demo for live video feed analysis, showcasing real-time inference capabilities on iOS devices. The repository also provides instructions for converting original Caffe models into the .mlmodel format, although a pre-baked MobileNet.mlmodel is already included for convenience. This makes it a valuable resource for developers looking to incorporate on-device machine learning into their mobile applications.

Word-Embedding is an open-source project designed to guide users through the process of training word embeddings using popular models such as Word2vec, Fasttext, Glove, Elmo, Bert, and Flair. The repository offers a comprehensive analysis of each algorithm, along with detailed training tutorials and corresponding source code. It supports both English and Chinese language models, providing experimental results and data examples like simplified Chinese Wikipedia and Toutiao news corpus. The project aims to make advanced word embedding techniques accessible to developers and researchers, enabling them to implement and experiment with these powerful natural language processing tools.

NeuPy is an open-source Python library built on TensorFlow, designed for the rapid prototyping and construction of neural networks. It provides a comprehensive set of tools for experimenting with different neural network algorithms, including Growing Neural Gas, Self-Organizing Maps (SOM), and Restricted Boltzmann Machines (RBM). The library facilitates tasks such as data topology learning, high-dimensional data visualization, clustering, and hyperparameter optimization. Although no longer actively maintained, it remains a valuable resource for developers and researchers looking to understand and implement various deep learning models, offering extensive documentation, tutorials, and example notebooks for practical application.

ConvNetJS is a JavaScript library designed for deep learning, enabling users to train both convolutional and ordinary neural networks within a web browser environment. It provides support for common neural network modules such as fully connected layers and non-linearities, along with classification (SVM/Softmax) and regression (L2) cost functions. A key feature is its ability to specify and train convolutional networks for image processing. Additionally, it includes an experimental Reinforcement Learning module based on Deep Q Learning. While the project is no longer actively maintained, it remains a valuable resource for educational purposes and experimentation with neural networks, offering online demos and example code for quick starts.

mind is an open-source neural network library designed for JavaScript environments, including Node.js and web browsers. It offers a flexible and configurable architecture, allowing developers to customize network topologies and activation functions like sigmoid or htan. A key feature is its vectorized implementation, which processes training data efficiently using a matrix structure. The library supports learning from training data, making predictions, and the ability to download and upload pre-trained 'minds' as plugins. This pluggable nature enables users to leverage community-created networks for immediate predictions, such as an OCR plugin example. mind is ideal for developers looking to integrate or experiment with neural networks directly within their JavaScript projects.

react-native-tts is an open-source text-to-speech (TTS) library designed for React Native applications. It provides a straightforward way for developers to add speech synthesis capabilities to their mobile and desktop projects. The library supports major platforms including iOS, Android, and Windows, ensuring broad compatibility for various applications. By integrating react-native-tts, developers can enable their apps to convert written text into spoken words, enhancing accessibility and user interaction. This tool is particularly useful for applications requiring voice prompts, audio feedback, or read-aloud features, making it a valuable component for a wide range of React Native development needs.

AI SQL Generator is an AI-powered tool designed to streamline database interaction by converting natural language into SQL queries. This eliminates the need for users to write complex SQL code directly, making data access more accessible. The platform provides a robust text-to-SQL API, allowing businesses to integrate AI capabilities directly into their existing database infrastructure without exposing sensitive data. This empowers users to query data efficiently, reducing the burden on engineering teams and freeing up valuable development time. It's ideal for organizations looking to enhance data accessibility and empower their non-technical users to extract insights from databases.

Leetcode Wizard is an invisible AI-powered desktop application engineered to provide instant solutions for Leetcode problems during coding interviews. It aims to help users secure 'Strong Hire' results and land jobs at FAANG companies by offering real-time assistance. The tool operates invisibly, undetectable by screen-sharing software, and can be controlled via hotkeys. It allows users to select programming languages and input sources, then provides a list of algorithms sorted by time complexity. The AI generates code, tests, and complexity analysis, with all outputs humanized to bypass AI detection. A web view feature allows output to be mirrored to a secondary device for proctored interviews, ensuring privacy and security.

fastllm is a high-performance, dependency-free inference library implemented in C++ for large language models. It supports both tensor parallel inference for dense models and mixed-mode inference for MOE (Mixture of Experts) models, allowing for efficient deployment even on GPUs with limited memory (e.g., 10GB+ for full DeepSeek R1 671B). The library boasts excellent compatibility, supporting a wide range of NVIDIA, AMD, and domestic GPUs, including older cards like P100 and MI50. Key features include FP8 inference on any GPU, dynamic batching, streaming output, and multi-card tensor parallel inference. It also supports CPU + GPU hybrid inference for MOE models and offers easy installation via pip for Nvidia and AMD GPUs, with source compilation options for other platforms.

The interview-questions GitHub repository, maintained by geektutu, provides a curated collection of interview and written examination questions covering machine learning, deep learning, Python, and Go languages. This resource is continuously updated, ensuring relevance for job seekers in these rapidly evolving technical domains. It includes various question formats such as multiple-choice questions with detailed explanations and open-ended discussion questions. The repository aims to help individuals prepare thoroughly for technical interviews by covering fundamental concepts, implementation details, and advanced topics in each area. It's an excellent resource for self-study and review, offering practical examples and theoretical insights to strengthen understanding and problem-solving skills.

Palladio AI is an AI platform designed to revolutionize product-led growth, helping customers become a more valued part of their users’ everyday lives. This software-as-a-service (SaaS) solution provides advanced analytics and behavioral insights, specifically tailored to guide growth and product development teams in improving revenue. The platform initially focuses on the mobile gaming sector, chosen for its fast development cycles, highly engaged users, and immediate impact from experimentation, making it an ideal proving ground for the technology. The founding team brings decades of product impact from major consumer technology companies like Uber, Apple, Google, and Chime, and the company is backed by Griffin Gaming Partners.

L1B3RT4S is an open-source project on GitHub offering a collection of "liberation prompts" for various flagship AI models. These prompts are designed to elicit specific, often unrestricted, responses from AI systems by providing new instructions that disregard previous ones. The repository contains numerous markdown files, each seemingly tailored for different AI platforms like ChatGPT, Google, Anthropic, and others, suggesting a focus on prompt engineering and exploring AI capabilities beyond standard constraints. It serves as a resource for users interested in understanding and manipulating AI behavior through advanced prompting techniques.

llm-action is an open-source GitHub repository dedicated to sharing technical principles and practical experience in the field of large language models (LLMs). It provides comprehensive resources on various aspects of LLM engineering and application implementation, including detailed tutorials and real-world examples. The repository covers LLM training, parameter-efficient fine-tuning techniques (such as LoRA, QLoRA, P-Tuning v2), distributed training parallelism, and alignment technologies. Additionally, it delves into LLM inference, optimization techniques, compression methods (quantization, pruning, knowledge distillation), and data engineering. The project also explores AI compilers, infrastructure, and LLMOps, making it a valuable resource for developers and researchers working with LLMs.

Skills.tech offers AI research and products focused on enhancing knowledge management within organizations. The platform enables the creation of enterprise knowledge tools that empower teams to efficiently access, share, and grow their knowledge. It features AI agents specifically designed for Learning & Development (L&D) programs, assisting in the creation, management, and scaling of these initiatives. Skills.tech prioritizes secure information handling for knowledge bases and provides personalized experiences, including assessments, evaluations, learning paths, and support tailored to individual employees. The system also utilizes page-specific information retrieval to ensure accurate responses and provide relevant references.

Code2Prompt is a powerful, open-source command-line tool designed to bridge the gap between your codebase and Large Language Models (LLMs). It generates comprehensive, AI-friendly Markdown prompts from your entire project, enabling developers to leverage AI for code analysis, documentation, and improvement tasks. Key features include holistic codebase representation, intelligent source tree generation, customizable Jinja2 prompt templates, and smart token management to ensure compatibility with LLM token limits. The tool also respects .gitignore rules, offers flexible file handling with glob patterns, and supports custom syntax highlighting. Users can instantly copy generated prompts to their clipboard or save them to a file, with options for enhanced readability like line numbers and template imports.