Skip to content
    Back to courses
    Learning path

    AI Architect: From AI Engineer to Expert

    15 lessons in two courses, from designing one AI system to running it in production. Figures first, interactive widgets, and case files on every page.

    Two courses, one path

    Intermediate teaches you to design the system. Advanced teaches you to operate, secure and govern it. Not sure where to start? The roadmap page has a 10-question self-assessment and the index of case files.

    TrackIntermediate ~13.5 hr8 lessons

    AI Architect (Intermediate): Design the System

    Learn to design an AI system: requirements, design reviews, diagrams, model routing, retrieval, agents, MCP, event-driven designs and evaluation.

    Lessons in this track
    Not started0 / 8 lessons
    TrackAdvanced ~15 hr7 lessons

    AI Architect (Advanced): Operate, Secure and Govern

    Run AI in production: LLMOps, reliability, cost, security, governance, cloud platforms and a capstone with two full designs, taught through case files.

    Lessons in this track
    Not started0 / 7 lessons

    What you will be able to do

    • Turn a vague AI feature request into requirements a test can check.
    • Draw one system as a context view, a component view and a runtime sequence.
    • Estimate the latency and cost of a request from its token counts and choose a routing policy.
    • Design retrieval, agent and event-driven systems, and say what happens when they fail.
    • Set up evaluation gates, tracing and release practices that catch regressions.
    • Design for outages, rate limits and cost, and threat-model prompt injection and tool abuse.
    • Turn regulation text into a registry, audit logging and human-oversight controls.

    How the lessons work

    • Every lesson opens with one overview figure, then explains with diagrams, tables and short text instead of long paragraphs.
    • Live widgets let you change the numbers: KV-cache size, retry multiplication, evaluation gates, cost per request and step-through walkthroughs.
    • Case files show each idea on one design problem, either from a published source or as a labelled composite scenario.
    • Technical lessons explain the mechanism inside the technology in three layers, so you can defend a design and not only copy one.

    Read the roadmap and case file index , or filter every case by industry, pattern and failure type in the Use Case Library .

    Questions

    Who is this course for?
    Engineers with two or more years of backend experience who have called an LLM API but have not designed a whole AI system, and the tech leads and engineering managers who sign off on such designs.
    Do I need both courses?
    Start with the Intermediate course unless you already write testable requirements, draw a system at several levels, and design retrieval, agents and evaluation. The roadmap page has a 10-question self-assessment that tells you where to start.
    Is it free?
    Yes. The first lesson of each course is open to everyone, and a free account unlocks the rest, tracks your progress and keeps your place.
    Are the case studies real companies?
    Some are based on an organisation's own published post or paper, with every fact attributed to its source, and are labelled Public. The others are composite scenarios, labelled "Composite scenario, not a real company", with illustrative numbers and stated assumptions.
    Is the governance material legal advice?
    No. It shows how to turn regulation text into engineering controls, quotes the official text with the date it was checked, and leaves legal decisions to counsel.
    How do I know the facts are right?
    Claims about products, papers and regulations are attributed to the primary source with the date they were checked, and the arithmetic is recomputed. The pages were drafted with AI assistance and checked against those sources; none carries the author's Reviewed badge yet.