Curated GenAI Resources

Community-curated feed of the most impactful AI research papers published daily. Includes summaries, discussions, and links to code. The best way to stay on top of what's actually moving the field forward.

Andrew Ng's weekly newsletter covering the most important AI news, research, and industry developments. Clear, concise analysis that cuts through the hype. Essential reading for anyone in the AI space.

Daily newsletter delivering the most important AI news in 5 minutes. Covers research, tools, product launches, and industry moves. Great signal-to-noise ratio for busy professionals.

Thoughtful, practitioner-focused coverage of LLMs, AI tools, and the developer experience. Simon is one of the most respected voices in the AI-for-developers space. Deep dives, not hot takes.

The AI assistant that started the generative AI revolution. GPT-4o powers the free tier with strong reasoning, code generation, and multi-modal capabilities. The most widely used AI tool in the world.

Anthropic's AI assistant, known for nuanced reasoning, long-context understanding (200K tokens), and strong safety properties. Excels at analysis, writing, and code. Free tier available with Claude Sonnet.

Google's multi-modal AI assistant with deep integration into Google Workspace. Strong at research, summarization, and analysis. Free tier includes Gemini Pro with generous daily limits.

AI-powered search engine that combines LLM reasoning with real-time web retrieval. Every answer includes citations. The best tool for research queries that need up-to-date, sourced information.

Google's AI research assistant that lets you upload documents and have grounded conversations with your own data. Generates audio summaries (podcasts) from your materials. Free and surprisingly powerful.

Anthropic's official CLI for agentic coding with Claude. Understands your entire codebase, executes commands, writes and edits files, and manages git workflows. The most capable AI coding agent available today.

AI-first code editor built on VS Code. Integrates multiple LLMs for code completion, chat, and multi-file editing. The 'Composer' feature can make changes across your entire codebase from a single prompt.

AI pair programmer that suggests code completions inline as you type. Powered by OpenAI Codex and GPT-4. Integrates deeply with VS Code and JetBrains IDEs. The most widely adopted AI coding tool in enterprise.

Open-source AI pair programming tool that works in your terminal. Connects to any LLM API and makes changes directly to your local git repository. Excels at multi-file edits and refactoring.

AI-powered IDE with a 'Flow' paradigm — the AI maintains awareness of your actions and proactively suggests next steps. Competitive free tier with multi-model support and deep codebase understanding.

The most popular framework for building LLM-powered applications. Provides abstractions for chains, agents, memory, retrieval, and tool use. Supports all major LLM providers. Large ecosystem of integrations and community examples.

Framework specialized for building RAG applications and data-augmented LLM systems. Excels at document ingestion, indexing, chunking strategies, and retrieval pipelines. The go-to choice when your primary use case is 'chat with your data'.

Direct API access to Claude models. Supports text generation, vision, tool use, and structured outputs. Known for clear documentation, predictable pricing, and strong developer experience. Essential for production AI systems.

Framework for evaluating RAG pipelines. Provides metrics like faithfulness, answer relevancy, context precision, and context recall. Essential for measuring and improving RAG system quality beyond vibes-based evaluation.

Open-source framework for building production-ready LLM applications with a pipeline-based architecture. Strong focus on RAG, document processing, and enterprise search. Component-based design makes it easy to swap in different models and retrievers.

Fully managed vector database purpose-built for AI applications. Offers serverless and pod-based deployment options. Known for simplicity, speed, and reliability. Free tier includes 100K vectors — enough for most prototypes.

Open-source vector database with built-in vectorization modules — it can generate embeddings for you using integrated model APIs. Supports hybrid search (vector + keyword), multi-tenancy, and GraphQL queries.

The simplest vector database to get started with. Runs in-memory or with persistent storage. Native Python and JavaScript clients. Perfect for prototyping and small-to-medium scale RAG applications.

PostgreSQL extension that adds vector similarity search to your existing Postgres database. No new infrastructure needed — if you already use Postgres, pgvector lets you add semantic search alongside your relational data. Production-proven at scale.

The GitHub of machine learning. Hosts 500K+ models, 100K+ datasets, and thousands of Spaces (demos). The central platform for discovering, sharing, and deploying open-source models. Essential for any ML practitioner.

Official library for parameter-efficient fine-tuning methods including LoRA, QLoRA, prefix tuning, and adapters. Integrates seamlessly with the Transformers library. The standard tool for fine-tuning LLMs on consumer hardware.

Transformer Reinforcement Learning library for training LLMs with RLHF, DPO, ORPO, and other alignment techniques. Provides trainers for SFT, reward modeling, and preference optimization. The standard tool for post-training alignment.

ML experiment tracking, model versioning, and dataset management platform. Track training runs, compare experiments, and collaborate with your team. The industry standard for ML experiment management. Generous free tier for individuals.

Library for 2-5x faster LLM fine-tuning with 80% less memory. Achieves speedups through custom Triton kernels and manual backpropagation. Compatible with Hugging Face Transformers. The fastest way to fine-tune open models.

High-throughput LLM inference engine using PagedAttention for efficient KV cache management. Achieves 2-4x better throughput than naive serving. Supports continuous batching, tensor parallelism, and speculative decoding. The standard for production LLM serving.

Hugging Face's production-ready LLM inference server. Optimized for Hugging Face models with features like continuous batching, quantization, speculative decoding, and grammar-constrained generation. Powers Hugging Face Inference Endpoints.

Enterprise-grade inference serving platform that supports any ML framework (TensorFlow, PyTorch, TensorRT, ONNX). Features dynamic batching, model ensembles, and GPU utilization optimization. The standard for multi-model serving at scale.

Scalable model serving framework built on Ray. Handles autoscaling, batching, multi-model composition, and GPU allocation. Excels at complex inference pipelines where multiple models need to work together. Used in production at OpenAI, Uber, and Spotify.

Run open-source LLMs locally on your machine with a single command. Supports LLaMA, Mistral, Gemma, and dozens more. Handles quantization, GPU acceleration, and model management. The easiest way to run LLMs locally for development and testing.

Free 1-hour courses taught by Andrew Ng and industry leaders. Topics include prompt engineering, LangChain, RAG, fine-tuning, and AI agents. Incredibly high quality for free content. New courses added weekly.

Free courses on prompt engineering, tool use, and building with Claude. Covers the Claude API, system prompts, function calling, and best practices. The most authoritative source for learning how to build with Claude effectively.

Comprehensive free course covering NLP with Transformers. Starts from tokenization and embeddings, progresses through fine-tuning and evaluation. Hands-on Colab notebooks throughout. The best free resource for understanding Transformers from the ground up.

Practical deep learning courses taught top-down — you build things first, then learn theory. Jeremy Howard's teaching style is legendary in the ML community. The 'Practical Deep Learning for Coders' course has launched thousands of ML careers.

Structured courses on building AI applications with LangChain, LangGraph, and LangSmith. Covers RAG, agents, evaluation, and production deployment. Ideal for developers building LLM-powered applications.

Andrew Ng's legendary ML course, available free on YouTube. Covers the mathematical foundations of machine learning: linear regression, SVMs, neural networks, and more. The gold standard for ML theory foundations.

Cloud GPU platform for AI training and inference. Offers on-demand and spot A100/H100 GPUs at competitive prices. Includes serverless GPU endpoints for inference. Popular in the open-source LLM community for fine-tuning.

Serverless cloud platform for running AI workloads. Define your environment in Python, and Modal handles containers, GPU scheduling, and scaling. Free tier includes $30/month in compute credits. Developer experience is exceptional.

Cloud and on-premise GPU infrastructure for AI. Offers H100 and A100 clusters for training and fine-tuning. Known for straightforward pricing and ML-focused tooling. Popular with research labs and startups.

GPU marketplace that connects renters with providers offering idle GPU capacity. Often 3-5x cheaper than major cloud providers. Best for batch workloads and fine-tuning where spot-like availability is acceptable.

Free hosted Jupyter notebooks with GPU access (T4). The simplest way to start experimenting with ML and LLMs without any local setup. Pro tier ($10/month) unlocks A100 GPUs and longer runtimes.