Resources

Google's Agent Development Kit (ADK) provides a comprehensive framework for building, testing, and deploying AI agents. It includes tooling for agent development with built-in functionality for reasoning, memory management, and multimodal interactions.

Amazon's smart speaker with advanced AI processing for voice recognition, natural language understanding, and smart home control, leveraging cloud-based and on-device machine learning.

A fully managed service that enables data scientists and developers to build, train, and deploy machine learning models quickly and easily. Includes hosted Jupyter notebooks, distributed training, and model monitoring.

AMD Instinct accelerators designed for AI and HPC workloads, including MI200 and MI300 series GPUs for data center applications.

AMD's newest accelerator combining GPU and CPU technology in a single package, designed for exascale computing and AI workloads with unprecedented efficiency.

AMD's most powerful accelerator for AI and HPC workloads featuring 153 billion transistors, 192GB HBM3 memory, and exceptional performance for LLM inference and training.

Free YouTube series building neural networks from scratch in Python, from micrograd to GPT-2. The most recommended LLM fundamentals course.

All-in-one desktop and Docker app for chatting with documents using RAG. Supports local and cloud LLMs with multi-user workspaces and agent tools.

Scalable deep learning framework that supports both imperative and symbolic programming, enabling flexible development and efficient deployment across devices from cloud infrastructure to mobile devices.

A dedicated neural network hardware component built into Apple's custom chips that accelerates machine learning tasks while using minimal power for enhanced on-device AI applications.

Apple's custom silicon neural engine integrated into its processors, designed to accelerate machine learning tasks for image recognition, natural language processing, and more.

A comprehensive library and hardware collection for interfacing various sensors with Arduino microcontrollers, perfect for robotics prototyping.

This paper introduces the Transformer, a novel neural network architecture based on a self-attention mechanism. The Transformer outperforms previous approaches on translation tasks while being more parallelizable.

Microsoft's framework for building multi-agent AI systems where multiple LLM agents collaborate, debate, and solve tasks through conversation.

Serverless compute service that runs code in response to events and automatically manages the compute resources, ideal for building AI-powered applications without managing servers.

Event-driven serverless compute platform that solves complex orchestration problems, with built-in AI integrations through Azure Cognitive Services and Azure Machine Learning.

Google's 2018 paper introducing BERT, the bidirectional encoder that revolutionized NLP transfer learning and dominated benchmarks for years.

A comprehensive library for interfacing with the BNO055 9-DOF absolute orientation sensor, providing easy access to orientation, acceleration, and magnetic data.

Deep learning framework developed by Berkeley AI Research, focused on expression, speed, and modularity. Particularly strong for computer vision applications and convolutional network models.

Lightweight, modular, and scalable deep learning framework built on the original Caffe. Designed for production deployment and mobile applications. Now part of PyTorch.

Cerebras Systems builds the largest chip ever made - the Wafer-Scale Engine - for AI training and inference at unprecedented scale.

Serverless GPU platform for deploying custom AI models. Supports any Python-based ML framework with cold starts under 5 seconds.

Google's paper showing that prompting LLMs to show step-by-step reasoning dramatically improves performance on math, logic, and commonsense tasks.

ChatGPT is a large language model from OpenAI that's optimized for conversation. It can answer questions, assist with writing, summarize content, translate languages, and more.

The AI-native open-source embedding database with cloud-hosted serverless options for production deployments.

Claude is an AI assistant created by Anthropic built to be helpful, harmless, and honest. It can understand and respond to nuanced instructions, provide thoughtful explanations, and assist with a wide variety of tasks.

Anthropic's latest mid-size model with exceptional reasoning, instruction following, and content creation capabilities. Designed to be more helpful, harmless, and honest than previous models.

Introduces CLIP, a neural network trained on a variety of image-text pairs that can be applied to any visual classification benchmark.

Run AI inference at the edge across Cloudflare's global network. Serverless, pay-per-use access to curated open-source models with ultra-low latency.

A large-scale object detection, segmentation, and captioning dataset with over 330K images, 1.5 million object instances, and 80 object categories.

Common Objects in Context (COCO) is a large-scale object detection, segmentation, and captioning dataset. The dataset contains over 330K images with 80 object categories.

Enterprise NLP API specializing in embeddings, reranking, and RAG. Command R models are optimized for enterprise search and agentic workflows.

Production-ready AI models via serverless API, including embeddings, text generation, and classification functions.

Cohere's enterprise-focused 104B open-weight LLM optimized for RAG, tool use, and multi-step agentic tasks in production environments.

Petabyte-scale open web crawl dataset updated monthly. The foundation for training most large language models including GPT and LLaMA.

A comprehensive course on computer vision fundamentals, covering image processing, object detection, and deep learning applications.

Open-source AI coding assistant for VS Code and JetBrains. Supports any LLM (local or cloud) for autocomplete, chat, and inline editing.

A Python library for designing and analyzing feedback control systems with tools for modeling, state estimation, and controller design for robotic systems.

Google's USB hardware accelerator for neural networks that adds fast ML inferencing capabilities to existing systems, optimized for TensorFlow Lite models.

Framework for orchestrating role-playing AI agents. Define agents with roles, goals, and tools, then have them collaborate as a crew to complete complex tasks.

DALL-E 3 is OpenAI's advanced text-to-image model that generates detailed and creative images from natural language descriptions. It can create almost any visual concept in various artistic styles with remarkable accuracy.

End-to-end serverless ML platform that enables building, training, and deploying ML models with automatic infrastructure management and MLOps capabilities.

A comprehensive course series that teaches the foundations of deep learning, how to build neural networks, and how to lead machine learning projects.

A comprehensive course on deep learning with PyTorch. This tutorial series covers everything from the basics to advanced topics in neural networks, computer vision, and natural language processing.

Collection of 1-2 hour practical AI courses taught by industry leaders (Andrew Ng, Harrison Chase, etc.) covering LLMs, RAG, agents, fine-tuning, and more.

Introduces DeepSeek-R1, a reasoning model trained with pure RL that matches OpenAI o1 performance while being fully open-source.

State-of-the-art open-source MoE model with 671B total parameters (37B active). Matches frontier closed models on coding, math, and reasoning benchmarks.

Serverless platform for JavaScript and TypeScript with built-in AI capabilities, featuring secure runtime and global distribution for deploying AI applications with low latency.

Next generation platform for object detection, segmentation, and other visual recognition tasks. Built on PyTorch, it's designed for flexibility and extensibility.

Open dataset of 3 trillion tokens used to train OLMo. Combines web text, code, scientific papers, books, and Wikipedia with full data provenance and filtering details.

Drake is a C++ toolbox for analyzing and optimizing the dynamics of robots and other mechanical systems, with a focus on control, planning, and verification.

Stanford's framework for algorithmically optimizing LLM prompts and weights. Replaces manual prompt engineering with programmatic pipelines and automatic optimization.

Sequence modeling toolkit that allows researchers and developers to train custom models for machine translation, text generation, language modeling, and other text generation tasks.

Top-down, practical deep learning course by Jeremy Howard. Covers vision, NLP, tabular data, and diffusion models with minimal math prerequisites.

Hugging Face's 15-trillion token high-quality web dataset derived from CommonCrawl with aggressive deduplication and filtering. Outperforms other web datasets on benchmarks.

Production-grade generative AI platform offering fast inference for open-source LLMs and multimodal models with serverless and dedicated deployments.

Introduces IO-aware exact attention that is 2-4x faster and uses 5-20x less memory than standard attention, enabling longer context windows in practice.

A powerful 3D simulation environment that provides realistic physics and sensor feedback for developing and testing robotic systems.

A powerful 3D simulation environment for autonomous robots that generates realistic sensor feedback, physically plausible interactions, and accurate dynamics. Ideal for testing robotics algorithms before real-world deployment.

Google's most capable multimodal AI model with a 1 million token context window, enabling unprecedented long-context understanding across text, code, audio, image, and video inputs.

Google's fastest and most efficient frontier model. Supports a 1M token context window, native multimodality, and real-time streaming at low cost.

Google's lightweight open-weight model family (1B–27B). Multimodal, multilingual across 140+ languages, and optimized to run on a single GPU or TPU.

A hands-on course covering the fundamentals of how generative AI works, and how to deploy LLMs responsibly.

High-level API for Apache MXNet, jointly developed by AWS and Microsoft. Makes deep learning easier with simple, concise API for defining models using a collection of pre-built, optimized neural network components.

A unified platform for building, training, and deploying machine learning models and AI applications at scale. Includes pre-trained APIs, AutoML for no-code model building, and custom training services.

Google's ASIC designed to run TensorFlow Lite models at the edge with high performance per watt, available in various form factors for IoT and embedded applications.

Gemini is Google's most capable AI model, built to be multimodal from the ground up. It can understand and reason about text, images, code, audio, and video, and generate content across these modalities.

Google's smart home devices with built-in machine learning capabilities for voice recognition, presence detection, thermostat optimization, and security monitoring.

Google's custom-designed Tensor SoC for Pixel phones, optimized for Google's ML algorithms to deliver improved photography, speech recognition, and other AI-powered features.

Google's custom-developed application-specific integrated circuits (ASICs) used to accelerate machine learning workloads, designed specifically for Google's TensorFlow framework.

OpenAI's flagship multimodal model processing text, audio, and images in real-time. Powers ChatGPT and the OpenAI API with the best overall performance.

The Intelligence Processing Unit (IPU) designed specifically for machine intelligence workloads, featuring massive parallelism and novel architecture for AI applications.

Grok is an AI assistant by xAI designed to answer questions with humor and personality. It features powerful reasoning capabilities and real-time knowledge about the world through web browsing functionality.

Ultra-fast LLM inference API powered by LPU hardware, offering some of the fastest token generation speeds available for models like Llama 3 and Mixtral.

A specialized Language Processing Unit (LPU) designed for ultra-fast AI inference with record-breaking speeds for large language model execution, featuring deterministic performance and low latency.

Open-source NLP framework for building production RAG pipelines, document search, and question answering systems with any LLM or retriever.

The largest hub for open machine learning datasets. Browse, search, and load 100,000+ datasets for NLP, vision, audio, and more with the datasets library.

Run thousands of open-source models via a simple API. No setup required — instantly access text, image, audio, and video models hosted by Hugging Face.

Free course covering Transformers, tokenizers, fine-tuning, and the entire HuggingFace ecosystem. Official and kept up to date.

Transformers provides thousands of pretrained models to perform tasks on different modalities such as text, vision, and audio. These models can be applied on text, images, audio and more.

Python library for structured outputs from LLMs using Pydantic models. Handles retries, validation, and streaming for reliable, type-safe LLM responses.

Intel's comprehensive portfolio of AI hardware solutions including CPUs, GPUs, FPGAs, and purpose-built AI accelerators for diverse workloads from edge to cloud.

Intel's high-performance AI accelerator for training and inference, featuring a specialized architecture optimized for deep learning with high-bandwidth memory and advanced compute capabilities.

Intel's purpose-built AI training processor designed for scalability, with integrated high-speed networking for distributed training of deep learning models.

Intel's low-power vision processing unit (VPU) designed for accelerating computer vision and deep neural network inferencing at the edge.

Intel's neural network processors optimized for deep learning in data centers, offering energy-efficient performance for AI inference workloads.

NVIDIA's physics-based simulation platform for developing, testing, and managing AI-based robots in photorealistic environments with real-time ray tracing.

Open-source ChatGPT alternative that runs 100% offline. Supports multiple local and remote AI engines with an extensible plugin system.

High-performance numerical computing and automatic differentiation library developed by Google Research. Combines Autograd and XLA for efficient ML research on accelerators.

Large-scale open dataset of 5.85 billion image-text pairs scraped from the internet, used to train Stable Diffusion and other vision-language models.

A serverless vector database for AI applications with automatic embedding generation and cost-effective storage solutions.

LangChain is a framework for developing applications powered by language models. It enables applications that are context-aware, reason, connect to other data sources, and integrate with agents for complex tasks.

OpenAI's landmark 2020 paper introducing GPT-3 (175B parameters) and demonstrating emergent in-context learning and few-shot prompting at scale.

Pythonic serverless platform for running and scaling AI workloads. Define your model as a Python class and deploy to GPUs in minutes with built-in autoscaling.

Llama 2 is Meta's next-generation open-source large language model, designed for dialogue and text generation. Available in various sizes from 7B to 70B parameters.

Run LLMs as a single executable file. Combines a model and runtime into one file that runs on any OS without installation — created by Mozilla.

Data framework for LLM applications. Specializes in ingesting, indexing, and querying private data with RAG pipelines and agentic workflows.

Practical course on building LLM-powered applications covering prompting, RAG, fine-tuning, evaluation, and deployment.

Desktop app for discovering, downloading, and running local LLMs on your computer. Supports GGUF models with a built-in chat UI and local server.

Introduces LoRA, a parameter-efficient fine-tuning method that reduces trainable parameters by 10,000x while matching full fine-tuning quality.

Stanford Machine Learning course by Andrew Ng covering supervised learning, unsupervised learning, and best practices for applying ML to robotics and AI applications.

A comprehensive course by Google that covers machine learning concepts, featuring interactive visualizations and real-world examples.

Introduces Mamba, a state space model (SSM) architecture that matches Transformer quality while scaling linearly with sequence length.

Official getting started guide for implementing Model Context Protocol servers and clients from Anthropic.

Collection of reference MCP server implementations for various services including filesystem, GitHub, Slack, Google Drive, and more.

Microsoft's comprehensive suite of AI services and tools for building intelligent apps. Includes pre-built APIs for vision, speech, language, and decision-making, plus custom model training capabilities.

Commercial-grade distributed deep learning toolkit. CNTK allows users to easily realize and combine popular model types and enables efficient implementation and execution of RNNs, CNNs, and feed-forward DNNs.

Midjourney is an AI image generation model that creates images from natural language descriptions. It excels at creating detailed artistic renderings, realistic images, and conceptual art based on text prompts.

Scalable, cloud-native vector database supporting trillion-scale data with managed serverless deployments.

Mistral 7B is a powerful open-source language model that matches or outperforms other models of similar size. It features improved attention mechanisms and efficient inference.

Presents Mixtral 8x7B, a sparse mixture-of-experts LLM that matches GPT-3.5 quality while only using 2 of 8 expert networks per token.

Open-source platform for managing the ML lifecycle including experiment tracking, model registry, and deployment. Works with any ML framework.

Cloud platform for running AI/ML workloads. Write Python functions, deploy instantly, and scale to thousands of GPUs with no infrastructure management.

Comprehensive guide to the Model Context Protocol (MCP), a standard for communication between AI models and applications. Learn about the protocol specification, implementation details, and best practices.

Anthropic official announcement introducing the Model Context Protocol (MCP), an open standard for connecting AI assistants to external data sources.

An advanced MPC framework for robotic systems that enables optimal control with constraints, ideal for complex dynamic systems.

MoveIt is the most widely used software for manipulation, incorporating the latest advances in motion planning, manipulation, 3D perception, kinematics, control and navigation.

A flexible navigation framework for ROS2 that provides path planning, obstacle avoidance, and localization capabilities for mobile robots.

A large-scale dataset for autonomous driving with 1000 scenes of 20 seconds each, with data from 6 cameras, 1 radar, and 5 lidars, including 3D bounding box annotations.

NVIDIA's flagship data center GPU for AI and HPC workloads with 80GB memory and multi-instance GPU technology for running multiple workloads simultaneously.

NVIDIA's most advanced Tensor Core GPU based on the Hopper architecture, designed for AI training and inference with transformers and large language models.

NVIDIA's small, powerful computer for embedded AI applications, delivers 472 GFLOPS for running modern AI algorithms with power efficiency in a compact form factor.

NVIDIA Jetson Orin is the world's most powerful AI computer for energy-efficient autonomous machines, delivering server-class performance for advanced robotics and edge AI applications.

NVIDIA's AI computer for autonomous machines featuring Volta GPU architecture and dedicated acceleration for deep learning, computer vision, and video processing.

NVIDIA's consumer and workstation GPUs featuring ray tracing and tensor cores, providing AI acceleration capabilities for content creation, gaming, and AI development.

Run large language models locally with a simple CLI and REST API. Supports Llama, Mistral, Gemma, Phi, and many other models out of the box.

Open Neural Network Exchange is an open format for representing machine learning models. ONNX defines a common set of operators and a common file format to enable model interoperability between frameworks.

Self-hosted, feature-rich web UI for LLMs. Compatible with Ollama and OpenAI-compatible APIs. Supports RAG, multimodal models, and multi-user management.

High-quality implementations of reinforcement learning algorithms. Baselines provides reference implementations to reproduce published results and serves as starting points for developing new approaches.

Toolkit for developing and comparing reinforcement learning algorithms. Gym provides a standardized set of environments for agent benchmarking and is compatible with any numerical computation library.

Human-generated, human-annotated assistant-style conversation corpus with 161,443 messages in 35 languages for training RLHF-based models.

A specialized collection of OpenCV-based tools and workflows designed specifically for robotics applications, from object detection to visual SLAM.

Comprehensive tutorial series on implementing computer vision for robotics applications using OpenCV, covering object detection, tracking, and environmental perception.

Unified API gateway for 100+ LLMs from OpenAI, Anthropic, Google, Meta, and more. Single integration, automatic fallbacks, and cost-based routing.

Library for structured text generation with guaranteed output formats. Enforces JSON schemas, regex patterns, and grammars directly at the token sampling level.

A fast, powerful, flexible and easy to use open source data analysis and manipulation tool built on top of Python.

A specialized toolkit and design guidelines for creating custom PCBs for robotics applications, including motor control, sensor integration, and power management.

Hugging Face library for fine-tuning LLMs with methods like LoRA, QLoRA, Prompt Tuning, and Prefix Tuning using minimal GPU memory.

An AI-powered search engine and conversational answer engine that combines multiple LLMs with real-time web search to provide accurate, up-to-date information with cited sources.

Microsoft's 14B parameter model that punches far above its weight on reasoning and STEM benchmarks, outperforming much larger models through data quality focus.

A comprehensive toolkitfor implementing and tuning PID controllers for various robotic systems, from simple robots to complex manipulators.

Managed vector database designed for machine learning applications, with real-time vector search and serverless operations.

A free course designed for people with coding experience who want to learn how to apply deep learning and machine learning to practical problems.

PyRobot is a Python framework for robotics research and benchmarking that provides a high-level interface for robot control and learning.

An open source machine learning framework that accelerates the path from research prototyping to production deployment.

Official PyTorch tutorials covering everything from basics to advanced topics in deep learning, with hands-on examples and code.

Vector database for similarity search with a focus on extended filtering support and serverless operations.

This paper presents a framework for learning agile legged locomotion skills for quadrupedal robots over challenging terrains.

A comprehensive solution combining Qualcomm's Hexagon processors, NPUs, and software tools to enable efficient AI processing across mobile and IoT devices for on-device machine learning.

Alibaba's open-weight model family (0.5B–72B) with strong multilingual, coding, and math capabilities. Top-performing open model in its size class.

Intel RealSense SDK provides tools and APIs for working with depth cameras, enabling robust 3D sensing capabilities for robotics applications.

Open reproduction of the LLaMA training dataset (1.2 trillion tokens) across 7 data sources. Enables fully open LLM pretraining research.

Run open-source machine learning models with a cloud API. Deploy and scale models instantly without managing infrastructure.

A curated collection of 100 diverse computer vision datasets spanning multiple domains like agriculture, retail, and industrial settings.

A comprehensive library for robotic vision applications, focusing on real-time performance and robustness inchallenging environments.

Robot Operating System 2 (ROS 2) is a set of software libraries and tools for building robot applications. From drivers to state-of-the-art algorithms, and with powerful developer tools.

An extension of ROS designed for industrial automation and robotics, providing packages and drivers for common industrial robot arms and peripherals.

A complete set of tools for autonomous robot navigation including mapping, localization, path planning, and obstacle avoidance capabilities for ROS2.

A comprehensive framework for implementing and testing various sensor fusion algorithms in ROS2 environment, optimized for robotics applications.

A reconfigurable dataflow architecture purpose-built for machine learning workloads, designed to scale from research to production with high performance and efficiency.

OpenAI's 2020 paper establishing power-law scaling relationships between model size, compute, data, and loss — the empirical foundation for scaling LLMs.

Simple and efficient tools for predictive data analysis. Built on NumPy, SciPy, and matplotlib, it's accessible to everybody and reusable in various contexts.

Introduces the Segment Anything Model (SAM), a promptable segmentation system trained on the largest segmentation dataset to date, capable of zero-shot transfer to new image distributions and tasks.

Microsoft's open-source SDK for integrating LLMs into applications. Supports plugins, planners, memory, and multi-agent patterns in Python, C#, and Java.

Comprehensive guide to sensors commonly used in robotics projects, including LiDAR, ultrasonic, infrared, and camera-based sensors with application examples and integration tutorials.

Educational resource designed to help anyone learn deep reinforcement learning. Provides clean implementations of key algorithms, educational exercises, documentation, and tutorials.

Stable Diffusion is an open-source text-to-image model that generates realistic images from text descriptions. It can be run locally on consumer hardware and supports various applications including inpainting, outpainting, and style transfer.

Open-source text-to-image generation model with significantly improved photorealism, prompt following, text rendering, and better understanding of complex scenes and instructions.

Self-instruct dataset of 52,000 instruction-following examples generated from GPT-3.5, used to fine-tune the original Alpaca model and spark the open instruction-tuning movement.

Stanford's flagship NLP course covering word vectors, RNNs, Transformers, LLMs, and modern NLP techniques. Free lecture videos and assignments available online.

Stanford's computer vision course covering CNNs, object detection, segmentation, and vision-language models. Lecture notes and assignments freely available.

Open-source app framework for Machine Learning and Data Science teams. Turn data scripts into shareable web apps in minutes with pure Python.

Serverless functions for building AI applications with Postgres integration, perfect for building LLM-powered apps with vector search capabilities.

An end-to-end open source platform for machine learning. Comprehensive, flexible ecosystem of tools, libraries, and community resources.

End-to-end platform for deploying production ML pipelines. TFX standardizes the components needed for ML engineering, including data validation, feature engineering, model analysis, and serving.

Lightweight version of TensorFlow for mobile and embedded devices. Enables on-device machine learning with low latency and small binary size.

The most popular self-hosted web UI for running local LLMs. Supports dozens of model formats with extensions for RAG, TTS, and character personas.

Large dataset of permissively licensed source code from GitHub (6.4TB) covering 358 programming languages. Used for training code LLMs like StarCoder.

A reading comprehension dataset consisting of questions posed on Wikipedia articles, where the answer is a segment of text from the corresponding article.

Serverless and dedicated AI inference platform with access to 100+ open-source models. Features fast inference and fine-tuning capabilities.

The InstructGPT paper introducing RLHF for aligning LLMs with human preferences, the technique behind ChatGPT.

Run Hugging Face Transformers directly in the browser using ONNX Runtime Web. No server required — supports text, vision, audio, and multimodal models.

Serverless vector database optimized for AI applications with pay-per-request pricing model.

A comprehensive toolkit for creating, editing, and validating Unified Robot Description Format (URDF) files for robot modeling and simulation.

Library for building AI-powered user interfaces with React Server Components and streaming responses from AI providers like OpenAI, Anthropic, and more.

High-throughput and memory-efficient LLM inference engine. Features PagedAttention for 24x higher throughput than HuggingFace Transformers.

A cloud-native, serverless vector search service with semantic search capabilities and GraphQL API.

An open-source robot simulator that provides a complete development environment for modeling, programming and simulating robots with accurate physics and 3D visualization.

Industry-standard MLOps platform for experiment tracking, dataset versioning, model evaluation, and hyperparameter optimization. Used by most AI labs.

Whisper is an automatic speech recognition (ASR) system trained on 680,000 hours of multilingual data, capable of transcription, translation, and language identification.

YOLOv8 is the latest version of the YOLO object detection model, offering improved accuracy and speed. It supports object detection, segmentation, classification, and tracking.