The Library

Reference, by topic

These are my personal notes from building my knowledge on AI and its various aspects. Whether you are a seasoned AI professional or just starting out, I hope something here helps.

Foundations

What is inside ChatGPT, Claude, and Gemini. The base layer: how large language models actually work, where the AI stack sits, and how AI systems remember.

Large Language Models

  • Everything about Large Language Models The technology behind ChatGPT, Claude, and Gemini. Open vs closed weights, and how to run models locally with tools like Ollama. Updated May 17, 2026

AI Memory

  • Everything about AI Memory How AI systems remember across conversations. Mem0, Zep, and Letta compared, plus where the new research is heading. Updated May 17, 2026

AI Engineering Stack

  • Everything about the AI Engineering Stack The full AI stack across seven layers, and which provider plays at each. Google, Microsoft, OpenAI, Anthropic, and the open-weight pieces underneath. Updated June 1, 2026

Protocols & Patterns

When you ask an AI to book a flight or check your inbox, something has to connect it to the real world. This is how.

Model Context Protocol

  • Everything about MCP The open standard that lets AI plug into tools and data. Anthropic shipped it, OpenAI and Microsoft adopted it, plus where the security holes are. Updated May 17, 2026

AI Agents

  • Everything about AI Agents AI that takes action, not just answers. How ChatGPT, Claude, and Microsoft Copilot approach it, and what it takes to ship one in production. Updated May 17, 2026

AI Harnesses

  • Everything about AI Harnesses The software wrapper that turns a stateless LLM into a working system. The autonomy spectrum (and why most "agents" are fake), the ReAct loop, and where LangGraph, Claude Code, and CrewAI sit on it. Updated June 7, 2026

Applications

AI that does not live in a chat window. Robots, self-driving cars, surgical systems.

Physical AI

Overview

  • Everything about Physical AI The full map. NVIDIA's Isaac stack, the LeCun JEPA competing thesis, the Jetson hardware ladder, autonomous vehicles, humanoids, surgical robotics, defense AI, and the regulatory landscape. Updated June 1, 2026

Four layer guides

  • Governance Who controls the robot, how safety is ensured, who is liable. EU AI Act, NIST AI RMF, ISO 10218:2025, NHTSA, FDA pathways, NVIDIA NemoClaw. Updated June 1, 2026
  • Hardware The body: sensors (RGB, depth, LIDAR), compute (NVIDIA Jetson from Orin Nano Super to Thor), actuators, structure. Verified products and prices. Updated June 1, 2026
  • Software The brain: runtime (JetPack, CUDA, TensorRT), perception (YOLO, Whisper, SAM3), decision (GR00T, Nav2), world models (Cosmos, V-JEPA), frameworks. Updated June 1, 2026
  • Systems What gets built: arms, mobile robots, autonomous vehicles (Tesla, Waymo, Zoox), humanoids (Figure, Agility, Unitree, 1X), drones, surgical robots. Updated June 1, 2026

Personal plans

  • Learning Plan A read-first four-phase study plan moving from foundations through the software stack and hardware to first builds with LeRobot and Isaac Sim. Updated May 17, 2026
  • Build Plan A sequenced hardware purchase + project plan. Three tiers (starter to humanoid) and six projects from Jetson hello-world through robotic arm to autonomous navigation. Updated May 17, 2026

Tools

How do developers actually build with AI? These are the tools that do the heavy lifting.

Claude Code Coming soon

Claude Desktop Coming soon

OpenAI Codex Coming soon

Microsoft Copilot Coming soon

Governance

Who decides what AI is allowed to do, and who is liable when it gets it wrong.

AI Governance Coming soon

The rules AI has to follow. NIST, the EU AI Act, and what it takes to ship responsibly.