Explains AI architecture at two layers — the internal Transformer model and the external system pattern (prompting, RAG, fine-tuning, agents) — and how to choose between them.

How embeddings and vectorization turn text into searchable meaning, the pipeline behind RAG and semantic search, and the practices that keep retrieval accurate.

Separates pre-training from fine-tuning, contrasts fine-tuning with RAG, and covers efficient methods (PEFT, LoRA) and alignment (RLHF, DPO).

The fundamentals of prompt engineering — clarity, context, and structure — and how it sits inside the broader discipline of context engineering.

The reusable prompting patterns — Chain of Thought, PTCF, and dual-prompting — that make complex LLM tasks more coherent, consistent, and controllable.