There’s a particular vocabulary that working software engineers absorb almost subconsciously over years — the half-noun, half-jargon shorthand that lets two engineers describe a complex system in three sentences. We’re going to RAG over the docs, return a JSON-Schema’d response, and stream tokens to the client via SSE. Each of those words has a specific technical meaning, a history, and a context in which it’s used correctly. Knowing them quickly is a non-trivial professional skill — every junior engineer eventually realizes that not knowing them is what’s blocking them from following design conversations.
This field guide is the explicit form of that vocabulary. It’s a 36-chapter reference, organized into five volumes, with each chapter a self-contained explanation of one concept — what it is, what problem it solves, what the working engineer needs to know about it. Built to be looked up rather than read end-to-end.
What it covers
Five volumes, thirty-six chapters. The chapter list is too long to enumerate inline, so here are the volume summaries:
Volume I — Foundations. Schema, JSON, API, Parser & Parsing, Regular Expressions, JSON Mode, JSON Schema, ReAct, Tailwind, BM25 Search.
Volume II — AI & Retrieval. Tokens & Context Window, Embeddings, Vector Search, RAG, Retrieval Strategies, Prompt Engineering, Function Calling, Streaming.
Volume III — Infrastructure. Container Orchestration, Microservices, Message Queues, Caching Layers, Database Choices, Auth Patterns, Observability, Rate Limiting.
Volume IV — Practice. Testing Strategies, CI/CD, Feature Flags, A/B Testing, Versioning, Migrations, Monitoring, Incident Response.
Volume V — Application. End-to-end worked examples that tie the previous four volumes together into recognizable systems.
Each chapter is roughly 100-300 words and includes diagrams or interactive figures where they help. The intent is a fast reference — when someone says “we’re using SSE for streaming,” you can look up SSE in fifteen seconds and know what’s being discussed.
The whole document is a 4,800-line standalone HTML page, so the table of contents is the way you navigate it. Sections are linkable so you can bookmark individual concepts.
Read it
← Back to Autonomy