Beginner
How LLMs work, prompting, embeddings and vector search.
- LLM fundamentals
- Tokenization
- Embeddings
- Sampling and decoding
- +5 more
Generative AI
36 lessons in 11 modules, about 18.5 hours in all. Choose a path by level, or browse the modules.
The Lowtide Sleeper. An overnight train from Harbour Terminus to the Far Capital. You are the new trainee while the crew learns to trust a new assistant system.
A path is a list of modules in a sensible order. Paths share lessons, so you can move between them.
How LLMs work, prompting, embeddings and vector search.
Retrieval-augmented generation and document AI.
Tools, agents, MCP, memory, multi-agent systems and safety.
Cost, serving, observability, evaluation, fine-tuning and system design.
Interview preparation across the course: fundamentals, RAG, agents, safety, production and system design.
Open any module to read its lessons in order.
Each character stands for one idea in the course. You will meet them as you go.
Tobias
Announcer
fluency and hallucination
Merrow
Records clerk
retrieval and RAG
Wick
Porter
tools and agents
Marta Voss
Engineer
cost, latency and evaluation
Fenn
Signalman
safety and guardrails
Build something that puts several lessons to work.
Read a lesson, check yourself, and come back to it later. The numbers below are counted from the course itself.