LLM & Agent Engineering Study Guide
A practical, no-fluff study guide for software engineers who want to get up to speed on LLM and agent engineering — whether that’s for building real systems at work or for interview prep.
How to use this guide
- Each topic in the left nav is a standalone page you can read in ~10 minutes.
- Every page follows the same shape: why it matters, core concepts, a mental model, interview questions with brief answer keys, and a short watch list of verified YouTube videos.
- Topics are ordered roughly from foundational to advanced. You can read them in order, or jump straight to the topic you need.
- Nothing here requires a backend, an API key, or a login — it’s just reading.
Roadmap
| # | Topic | Interview |
|---|---|---|
| 01 | LLM basics for SWEs | Occasional |
| 02 | Tokens, context windows, and cost | 🎯 Frequent |
| 03 | Prompting & structured outputs | 🎯 Frequent |
| 04 | Tools / function calling | 🎯 Frequent |
| 05 | Agents & planning loops | 🎯 Frequent |
| 06 | Memory (short / long / episodic) | 🎯 Frequent |
| 07 | Context management & compaction | 🎯 Frequent |
| 08 | State machines & workflows | Occasional |
| 09 | Multi-agent / inter-agent communication | 🎯 Frequent |
| 10 | RAG fundamentals | 🎯 Frequent |
| 11 | Advanced RAG | 🎯 Frequent |
| 12 | Embeddings & vector DBs | 🎯 Frequent |
| 13 | Evaluation & observability | 🎯 Frequent |
| 14 | Post-training (SFT, RLHF/RLAIF, preference tuning) | Background |
| 15 | Safety, guardrails, trust | Occasional |
| 16 | Production patterns | 🎯 Frequent |
Contributing
See CONTRIBUTING.md. Short version: keep topics tight, verify every link yourself, don’t invent scope.