Coding & Development

EzAudio ControlNet is an innovative AI tool designed for generating new audio content. Users can provide a text description outlining the desired audio characteristics and upload a reference audio file to guide the generation process. The application then creates a new audio clip that incorporates elements from both the text prompt and the reference audio, offering a unique way to control audio output. Built with Gradio and hosted on Hugging Face, this tool is accessible via the web and operates under an MIT license, making it a free and open-source solution for audio creation and manipulation.

FLUX.1 [dev]-De-Distill is an AI tool hosted on Hugging Face, specifically designed for AI model development and machine learning research. It caters to the needs of AI researchers and developers, providing a platform for their work. The tool operates under the MIT license, promoting open access and collaboration within the AI community. Currently, the Space is paused, and users interested in utilizing it are directed to the community tab to request its restart from the author(s). This indicates a community-driven approach to its availability and maintenance.

FlexTok is a demo for flexible sequence length autoencoding, developed by EPFL-VILAB and available as a Hugging Face Space. This tool allows users to upload an image and generate various reconstructions by manipulating token sequences of different lengths. Users can customize parameters such as the seed, timesteps, and resolution to explore different levels of detail and output variations. Built with Gradio and licensed under Apache-2.0, FlexTok is designed for researchers and developers interested in experimenting with sequence modeling and understanding its effects on image reconstruction. It provides a hands-on platform to observe how changes in token sequence length and other settings influence the generated output.

gemma-3-270m is an AI chatbot that leverages the Gemma 3 (270M) language model, running efficiently on Ollama with just a single-core CPU. This tool is designed for users who need to experiment with and deploy AI models even with limited computational resources. It supports both the google/gemma-3-270m and google/gemma-3-270m-it models, providing flexibility for different applications. Users can input text prompts and receive generated responses, with options to customize output parameters such as context length, temperature, and repetition penalty. The platform is hosted as a Hugging Face Space, making it accessible for testing and development.

ICLR2024 Papers is a Hugging Face Space designed to help users navigate the research papers from the ICLR 2024 conference. This tool enables efficient searching of papers by title or abstract, and offers filtering capabilities based on paper type and associated links. Users can also claim authorship of their papers directly through the platform. It presents a comprehensive table of papers, complete with essential details and direct links, making it a valuable resource for academics and researchers looking to explore the latest advancements in AI and machine learning presented at ICLR 2024.

SecTools is a platform dedicated to providing practical and straightforward cybersecurity tools. It caters to both beginners looking to understand security concepts and experienced practitioners needing efficient utilities. The core philosophy of SecTools is to offer functional solutions without the typical overhead of sign-ups or excessive features, ensuring a direct and effective user experience. The tools are designed to address real-world cybersecurity challenges, making them valuable for various security-related tasks. This approach emphasizes accessibility and utility, allowing users to quickly leverage cybersecurity capabilities without friction.

LockMemo is a robust security tool designed for creating and sharing private, code-sealed notes with granular access control. Users can write a note, lock it with a code, and share a link, maintaining full control over who can read the content. It integrates with email services like Gmail, allowing users to send sensitive messages without attaching files directly. This feature helps prevent accidental sharing of confidential information and allows for setting expiration times on notes. Access can be revoked or managed at any time, and notes are protected with server-side encryption. LockMemo offers a simple three-step workflow: create and seal, share the link, and validate access, with options for manual approval or auto-acceptance after 48 hours.

Aerobotics7 develops an end-to-end technology platform for subsurface threat detection and mapping, specifically targeting landmines and unexploded ordnance (UXOs). Founded in 2016 by Harshwardhan Zala, the platform, known as EAGLE A7, aims to provide a safer, faster, and more accurate alternative to traditional detection methods such as manual probing and conventional ground-penetrating radar. It is designed to detect modern, plastic-based composite landmines and explosives with greater accuracy, significantly reducing human exposure to hazardous areas. Aerobotics7 works with governments, international organizations, and humanitarian agencies to address the global threat of buried landmines.

MACE (Mobile AI Compute Engine) is an open-source deep learning inference framework specifically designed for mobile heterogeneous computing platforms. It optimizes AI model deployment on Android, iOS, Linux, and Windows devices by focusing on performance, power consumption, responsiveness, and memory usage. Key optimizations include NEON, OpenCL, and Hexagon acceleration, Winograd algorithm for convolution, and chip-dependent power options. MACE also prioritizes model protection through techniques like converting models to C++ code and literal obfuscations. It supports popular model formats such as TensorFlow, Caffe, and ONNX, making it a versatile tool for developers working with mobile AI applications.

notte is a robust framework designed for rapidly building and deploying reliable web automation agents. It offers a full-stack solution that integrates AI agents with traditional scripting, allowing users to leverage AI for complex, non-deterministic tasks while using scripting for predictable parts. This hybrid approach significantly reduces costs by over 50% and enhances reliability. notte provides essential tools for developing, deploying, and scaling agents and web automations through a single API. Key features include an open-source core for running web agents, structured output with Pydantic models, and advanced site interactions. The API service further offers stealth browser sessions with CAPTCHA solving, proxies, and anti-detection capabilities, along with enterprise-grade credential management via Secrets Vaults and Digital Personas for automated 2FA.

Perception Models is a comprehensive repository offering state-of-the-art AI models for image, video, and audio perception. It features the Perception Encoder (PE) for robust encoding across various modalities, including core vision-language tasks, LLM-aligned vision-language modeling, and spatially-tuned dense prediction. Additionally, it provides the Perception Language Model (PLM) for decoding, facilitating research in vision-language modeling with open and reproducible models. The repository also includes PE Audio-Visual and PE Audio-Frame models, expanding its capabilities to joint audio-visual embedding and audio event localization. With extensive benchmarks and clear getting started guides, Perception Models is an invaluable resource for developers and researchers working on advanced multimodal AI applications.

PyTorchVideo is a deep learning library specifically designed to accelerate video understanding research. Built using PyTorch, it offers a comprehensive set of reusable, modular, and efficient components for developing video analysis models. Key features include a reproducible model zoo with state-of-the-art pretrained video models and benchmarks, extensive data loaders supporting various datasets, and video-focused fast components that enable accelerated inference on hardware. The library supports different deep learning video components like video models, video datasets, and video-specific transforms, making it easy to integrate with the broader PyTorch ecosystem. It is ideal for researchers and engineers working on advanced video-related AI applications.

WebLLM Playground is a Hugging Face Space that hosts a React application built with TypeScript, designed for interactive experimentation with language models. Users can access this playground directly in their web browser without needing to provide any initial input, making it highly accessible for quick exploration. The tool allows individuals to test prompts and understand the capabilities of various language models within a user-friendly interface. It is particularly suitable for those in research, development, or educational settings who need a straightforward way to interact with and learn about LLMs.

Swin-Unet is an open-source code repository implementing a Unet-like pure transformer for medical image segmentation, as detailed in the ECCVW 2022 paper "Swin-Unet: Unet-like Pure Transformer for Medical Image Segmentation." The repository offers updated reproducibility information, including instructions for downloading pre-trained Swin Transformer models and preparing datasets. It outlines the environment setup with Python 3.7 and provides scripts for training and testing the model on datasets like Synapse, BTCV, and ACDC. The project emphasizes the importance of pre-training for pure transformer models, with both encoder and decoder initialized with pre-trained weights to achieve optimal segmentation results.

VESSL AI offers a Liquid AI Infrastructure and Persistent GPU Cloud solution, providing on-demand access to a range of GPUs including A100, H100, H200, B200, GB200, and B300. Designed for researchers, AI startups, and enterprise AI teams, it allows users to spin up resources in minutes and scale on demand, paying only for what they use. The platform supports multi-node training, parallel jobs, and persistent workspaces, aiming to save users up to 80% compared to hyperscalers. It features options for spot, on-demand, and reserved capacity, with multi-cloud failover built-in and 24/7 platform monitoring. VESSL AI is SOC 2 Type II Certified and ISO 27001 compliant, ensuring secure and reliable operations for critical AI workloads.

yobulkdev is an open-source and AI-driven data onboarding platform designed as a free alternative to Flatfile.com. It simplifies data exchange for businesses that primarily use CSV and Excel files. The platform enables users to create a CSV import button, significantly accelerating the data importing process. Key features include no-code template creation, smart auto-matching between CSV and template columns, custom validation rule settings, and a delightful data review experience. It is scalable through streaming, capable of importing CSVs up to 1GB, and integrates GPT-3 for AI-based auto-suggestion and error correction. YoBulk aims to provide developers with a "bring your own validations" and "bring your own database" experience, ensuring data security and long-term maintainability for enterprises.

Pangolin is a zero-trust remote access platform designed for IT/OT, IoT, and engineering teams, offering a secure and identity-based alternative to traditional VPNs. It provides unified access to entire infrastructures, connecting via peer-to-peer tunnels or clientless browser access across on-prem, cloud, and edge environments. Key features include an easy-to-deploy connector behind any firewall, user authentication with existing identities, zero-trust access to specific applications, and enforcement of identity and context-aware rules. Pangolin is built to be easy to deploy and scale, verifying identity and context at every step to deliver secure access without the friction often associated with VPNs. It supports various platforms including MacOS, iOS, Windows, Linux, and Android, and offers both cloud and self-hosted options.

Open Neuromorphic is a global community dedicated to advancing brain-inspired AI and hardware through education, research, and open-source collaboration. It offers a comprehensive Computing Hub with guides for neuromorphic hardware and software, including SNN frameworks and event-based data tools. The platform facilitates community-driven projects through its Mission Board, hosts a peer review program for open and reproducible research, and organizes educational events like Hacking Hours, Student Talks, and expert-led Workshops. Members can explore resources, get involved in initiatives, and contribute to a collective knowledge base through blogs and presentations, fostering innovation in neuromorphic computing.

aerosolve is a machine learning library developed by Airbnb, designed with a strong emphasis on human interpretability and user-friendliness. It stands out from other ML libraries through its unique thrift-based feature representation, which supports pairwise ranking loss and single-context multiple-item representation. The library also features a powerful feature transform language, allowing users extensive control over feature engineering and rapid iteration. It is particularly well-suited for sparse, interpretable features commonly found in search or pricing applications, rather than dense, non-interpretable data like raw pixels. aerosolve includes debuggable models such as linear and spline models, facilitating insight into model behavior and feature relationships.

are-we-learning-yet is an open-source project dedicated to cataloging and evaluating the readiness of Rust for machine learning applications. Inspired by the 'Are We Web Yet?' initiative, this resource provides a curated list of Rust ML crates, along with metadata fetched from crates.io and the GitHub API. The project includes a scraper tool that generates scores for ordering crates and caches data to optimize site generation. It welcomes community contributions for adding missing crates, providing additional resources, and improving content, making it a collaborative effort to track the evolving Rust ML ecosystem.

AutoGL is an open-source AutoML framework and toolkit specifically designed for machine learning on graphs. It enables researchers and developers to easily and quickly conduct automated machine learning tasks on graph datasets. The framework supports various graph-based machine learning tasks through its auto solver, which integrates five main modules: auto feature engineer, neural architecture search (NAS), auto model, hyperparameter optimization (HPO), and auto ensemble. AutoGL is compatible with popular graph libraries like PyTorch Geometric (PyG) and Deep Graph Library (DGL), supporting tasks such as node classification, link prediction, and graph classification. It also serves as a flexible framework for implementing and testing custom AutoML or graph-based machine learning models.

Awesome AI SDKs is a curated database of essential SDKs, frameworks, libraries, and tools specifically designed for the development, monitoring, debugging, and deployment of autonomous AI agents. This resource aims to be a valuable starting point for developers and teams looking to build sophisticated AI agent solutions. The list, while not exhaustive, is actively maintained and encourages community contributions via pull requests. It is backed by the team at e2b, who are building an operating system for AI agents, providing a suite of tools, environments, SDKs, and APIs that are tech-stack agnostic.

awesome-game-ai is an open-source repository offering a curated collection of resources for game AI, specifically focusing on multi-agent reinforcement learning. It covers both perfect and imperfect information games, categorizing materials by game type. The repository includes open-source projects, review papers, research papers, conference information, and competitions related to game AI. It highlights advancements in games like Starcraft, Dota 2, Go, Chess, and various card games, providing valuable insights for researchers and developers in the field. Contributions to the list are welcomed via pull requests.

Plumerai develops software building blocks that enable customers to embed production-worthy AI inside their products, focusing on the full AI stack from data to hardware optimizations. Their people detection AI is highly accurate and resource-efficient, running on nearly any CPU, including $1 microcontrollers, with a memory footprint of just 1MB. The company offers a complete software solution for smart home cameras, including familiar face identification, stranger identification, people detection, vehicle detection, and advanced motion detection. This AI software is deployed on major camera SOC and cloud platforms, ensuring compliance with privacy laws like GDPR, CCPA, and BIPA. Plumerai's technology eliminates false alarms from traditional smart home cameras, providing relevant notifications and enhancing user experience.