Introduction
Welcome to AI Summary Hub — your single source of truth for modern AI concepts.
This hub is built for deep knowledge: each topic gives you clear definitions, how it works (with diagrams and code where useful), and links to official documentation and codelabs so you can go from understanding to building.
What you'll find here
This wiki covers 50+ topics across:
- Fundamentals — Machine learning, deep learning, neural networks
- Transformers & LLMs — Architecture, BERT, GPT, fine-tuning, prompt engineering, streaming
- RAG — Retrieval-augmented generation, vector databases, embeddings
- Agents & subagents — AI agents, multi-agent systems, hierarchies
- Reasoning patterns — Chain-of-thought, tree-of-thoughts, ReAct, RDD (retrieval-decision-design)
- Spec-driven development — Building AI systems from specifications
- Fields — NLP, computer vision, speech, robotics, multimodal AI
- Safety, ethics, evaluation — AI safety, bias, explainability, benchmarks
- Infrastructure & deployment — Local inference, edge reasoning, model compression, quantization
- Tools — Hugging Face, LangChain, Cursor, Claude Code, Antigravity, Kiro, PyTorch, TensorFlow
- Case studies — ChatGPT, DALL·E, Claude, Gemini, BART, Grok, DeepSeek, Qwen
Each topic includes definitions, examples (code and diagrams), pros/cons, benchmarks, and external documentation links to official docs, codelabs, and papers.
Learn more elsewhere
- Google AI for Developers — Gemini, APIs, and guides
- Hugging Face NLP Course — From transformers to LLMs
- From Prototypes to Agents with ADK (Codelab) — Build your first agent with Google's Agent Development Kit
Getting started
- New to AI? Start with Fundamentals and Neural Networks.
- Building with LLMs? See LLMs, RAG, and Agents.
- Exploring reasoning? Check Reasoning Patterns and RDD.
- Running models locally or at the edge? See Local inference and Edge reasoning.
- Multimodal (text + image, etc.)? See Multimodal AI.
Use the sidebar to browse all topics or the search bar to find specific concepts.