Generative AI
SearchStart here

Modules

11 modules, 36 lessons, about 18 h 40 min in all. Take them in order, or jump to the level you need.

Beginner3 modules

  • Module 1Beginner

    LLM foundations

    Dusk at Harbour Terminus, and the Engine Room is warming up.

    • Explain how an LLM generates text: attention, the KV-cache, tokenization and sampling.
    • Choose sensible sampling settings and compare open and closed models.

    3 lessons · 1 h 25 min

  • Module 2Beginner

    Prompting and structured output

    Evening in the Dispatch Office.

    • Pick between zero-shot, few-shot and chain-of-thought prompting for a task.
    • Get reliable structured output and know when a reasoning model is worth its cost.

    2 lessons · 55 min

  • Module 3Beginner

    Embeddings and vector search

    At the first junction the Sorting Depot has no labels, only zones painted on the floor, and parcels are sorted by what they are like.

    • Explain embeddings, similarity metrics and approximate nearest-neighbour search.
    • Choose a vector store for a given scale.

    2 lessons · 1 h

Intermediate2 modules

  • Module 4Intermediate

    Retrieval-augmented generation

    Late evening, and Tobias has to announce from the real timetable.

    • Explain every stage of a RAG pipeline and its trade-offs.
    • Diagnose a failing RAG system by separating retrieval from generation, and choose a RAG variant for a scenario.

    6 lessons · 3 h

  • Module 5Intermediate

    Multimodal and document AI

    The border station, a little after nine in the evening, and the Ticket Hall and Baggage Scan deal in stamped, smudged and handwritten paper.

    • Choose a parsing and extraction approach for PDFs, tables and images, and evaluate it.

    1 lesson · 35 min

Advanced3 modules

  • Module 6Advanced

    Tools and agents

    Night has fallen and the luggage needs moving.

    • Explain how tool calling works and design a tool-using agent.
    • Use MCP to connect tools and know its security risks.

    3 lessons · 1 h 35 min

  • Module 7Advanced

    Context, memory and multi-agent systems

    Around eleven the crew changes over, and the Guard's Logbook is the only thing that carries the night from one shift to the next.

    • Apply context engineering and design agent memory.
    • Choose between a workflow, a single agent and multiple agents, and say how you would evaluate it.

    3 lessons · 1 h 20 min

  • Module 8Advanced

    Safety and guardrails

    At 11:40 pm the train reaches the Signal Box.

    • Design a layered defence against prompt injection that does not rely on the system prompt.

    2 lessons · 50 min

Production3 modules

  • Module 9Production

    Production: cost, serving and observability

    Between one and three in the morning the Sleeper climbs the Mountain Pass in a storm, with a full train and a timetable that changed at…

    • Measure first, then reduce cost and latency, and state each lever's quality cost.
    • Deploy and observe a RAG system with an eval gate and rollback.

    5 lessons · 2 h 30 min

  • Module 10Production

    Evaluation and model adaptation

    Before dawn the Sleeper rests in the Repair Siding.

    • Build and interpret evaluations of LLM systems.
    • Decide between prompting, RAG and fine-tuning, and know how LoRA, QLoRA and DPO work.

    4 lessons · 1 h 45 min

  • Module 11Production

    GenAI system design

    Dawn at the Far Capital, and the Network Office hands you a blank sheet: design the next system.

    • Produce a design with requirements, estimates, an evaluation plan and failure modes, raising trade-offs unprompted.

    5 lessons · 3 h 45 min