This page maps the dependencies between chapters. Use it to plan your reading order if you want to skip ahead to a specific topic.
The Recommended Sequential Path¶
Ch 1 (AI at a Glance)
↓
Ch 2 (LLM Foundations)
↓
Ch 3 (LLM Providers) ──→ Ch 4 (Local LLMs)
↓ ↓
Ch 5 (Vibe Coding) ←────────┘
↓
Ch 6 (RAG First Principles)
↓
Ch 7 (Vector DBs) ──→ Ch 8 (Structured Data) ──→ Ch 9 (Multimedia)
↓
Ch 10-12 (LlamaIndex → LangChain → LangGraph)
↓
Ch 13 (Fine-Tuning)
↓
Ch 14-17 (Single Agent → Multi-Agent → Autonomous → Tool Agents/MCP)
↓
Ch 18 (Deployment)What You Can Skip To¶
If you already have experience with a topic, here is what each chapter actually requires:
| Chapter | Prerequisites | Can skip if you know... |
|---|---|---|
| Ch 1: AI at a Glance | None | How LLMs are trained (pre-training, SFT, RLHF) |
| Ch 2: LLM Foundations | None | Tokenization, embeddings, attention, transformers |
| Ch 3: LLM Providers | Ch 2 | How to call OpenAI/Gemini APIs |
| Ch 4: Local LLMs | Ch 3 | How to run LM Studio or Ollama |
| Ch 5: Vibe Coding | Ch 3 | How to use Cursor, Copilot, or Claude Code |
| Ch 6: RAG First Principles | Ch 2, Ch 3 | Chunking, embedding, cosine similarity, prompt stuffing |
| Ch 7: RAG with Vector DBs | Ch 6 | ChromaDB, Pinecone, FAISS |
| Ch 8: RAG with Structured Data | Ch 6 | Text-to-SQL, Cypher, schema-as-context |
| Ch 9: RAG with Multimedia | Ch 6 | CLIP, image/video embeddings |
| Ch 10: LlamaIndex | Ch 6 | LlamaIndex SimpleDirectoryReader, VectorStoreIndex |
| Ch 11: LangChain | Ch 6 | LCEL pipe syntax, LangChain chains and agents |
| Ch 12: LangGraph | Ch 11 | StateGraph, conditional edges, graph loops |
| Ch 13: Fine-Tuning | Ch 2, Ch 3 | LoRA, QLoRA, SFTTrainer |
| Ch 14: Single Agent | Ch 3 | LLM + tools pattern, system prompts |
| Ch 15: Multi-Agent | Ch 14 | Agent handoff, LLM-as-a-judge |
| Ch 16: Autonomous Agents | Ch 14 | Plan-Execute-Synthesize-Reflect pattern |
| Ch 17: Tool Agents and MCP | Ch 14 | MCP protocol, JSON-RPC, Cloud Run containers |
| Ch 18: Deployment | Ch 17 | Cloud Run deployment, Dify workflows |
Three Fast Tracks¶
“I just want to build a document Q&A app”: Ch 1 → Ch 2 → Ch 3 → Ch 6 → Ch 10 (LlamaIndex)
“I want to understand agents”: Ch 1 → Ch 2 → Ch 3 → Ch 14 → Ch 15 → Ch 16 → Ch 17
“I want to fine-tune a model”: Ch 1 → Ch 2 → Ch 3 → Ch 13