AI Engineering
How AI-powered software is built, from protocols to patterns.
MCP (the Model Context Protocol — how AI connects to tools, data, and workflows), agents (how AI uses tools and reasons through multi-step tasks), context management (the memory problem — why AI forgets and how to fix it), prompt engineering (structuring AI interactions for reliable output), evaluation (measuring whether AI output is actually good), and orchestration (coordinating multiple AI agents on one task).
This is the engineering discipline that didn't exist five years ago. Every section in this collection — systems, networking, databases — is knowledge the AI uses. This section is about building the systems that USE that knowledge through AI.
Lessons
1How MCP Works — The Universal Connector for AI2How MCP Servers Work — Exposing Tools and Data to AI3How MCP Clients Work — Connecting AI to Servers4How MCP Tools Work — Executable Functions for AI5How MCP Resources Work — Data Sources for AI Context6How MCP Transports Work — stdio, HTTP, and Message Framing7How AI Agents Work — Tool Use, Reasoning, and the Action Loop8How Context Management Works — Why AI Forgets and How to Fix It9AI Engineering FAQ