BEGINNER
How large language models work, how to prompt them effectively, and how to build production applications on top of them — from transformer intuition and tokenisation to RAG pipelines, tool use, and AI agent design.
What large language models are, how they turn text into tokens, the intuition behind the transformer architecture, and the boundaries of what they can and cannot do.
Techniques for writing instructions that reliably get the output you want — from zero-shot and chain-of-thought prompting to system prompts, personas, and defences against hallucinations and prompt injection.
The messages format, streaming, tool use, function calling, and the practical mechanics of cost, latency, and context-window management for production applications.
How embeddings encode meaning as vectors, how vector databases enable semantic search, and how retrieval-augmented generation (RAG) grounds LLM responses in real data.
How LLM agents plan and use tools across multiple steps, the fundamentals of AI safety and alignment, and a capstone challenge to design a complete AI assistant.