Learning path
From curious to AI solution designer
Start with no AI background. Three stages, one destination: each course makes you more capable of the next. Courses marked planned are still being written, so this path grows over time.
- 1Aware
- 2Practitioner
- 3Designer
- 4Solution designer
11 available now4 planned0 of 15 complete
Stage 1
Understand
You can explain AI and use it well.
0 of 4 complete
What AI is and is not · 7 lessons · 12 hr
After this you can: Explain AI, machine learning and LLMs to a colleague, and spot the hype.
Prompting, limits, tools · 7 lessons · 11 hr 30 min
After this you can: Get reliable results from chat models and know where they fail.
Notebooks, APIs, JSON · 7 lessons · 13 hr 30 min
After this you can: Run notebooks, call APIs and handle data well enough to build.
Train, test, evaluate · 7 lessons · 12 hr
After this you can: Know when classic machine learning beats an LLM, and how to judge a model.
Stage 2
Build
You can ship a small AI solution safely.
0 of 4 complete
Structured output, tool calls · 7 lessons · 13 hr 30 min
After this you can: Build small apps that call a model, return structured data and use tools.
- 6
Chat with your documents (RAG)
PlannedRetrieval from scratch
After this you can: Ship a document question-and-answer app and explain how it finds answers.
- 7
Evaluate and use AI responsibly
PlannedEvals, privacy, bias
After this you can: Test an AI feature, protect data, and say when not to ship.
Review, test, scan · 7 lessons · 15 hr
After this you can: Review, test and scan AI-written code before it reaches production.
Stage 3
Design
You can design agent-based systems.
0 of 6 complete
- 9
Agents fundamentals
PlannedLoops, tools, workflows
After this you can: Tell a workflow from an agent, and build a simple tool-using agent.
Agents that use the web · 7 lessons · 15 hr 15 min
After this you can: Give an agent a browser and keep it reliable, safe and polite.
Retrieval over connected data · 7 lessons · 14 hr 15 min
After this you can: Answer connected, multi-hop questions with a knowledge graph.
Agents that remember · 7 lessons · 13 hr 45 min
After this you can: Design memory that stays accurate, private and affordable.
Agents that fit your repo · 7 lessons · 15 hr 15 min
After this you can: Make coding agents effective in a real codebase, and measure it.
Evaluate, trace, secure, release · 7 lessons · 16 hr 45 min
After this you can: Evaluate, trace, secure and release an agent you can trust.
Final course · Level 4, solution designer
Capstone: AI solution designer
Pick a real problem, frame it, choose an architecture, weigh risk and cost, ship it, and explain the result. This is where you start to think in AI.
Planned
Specialize after the path (pick any)
A domain example · 7 lessons · 15 hr 15 min
After this you can: Apply the path to regulated, high-stakes workflows.
Expert · 7 lessons · 19 hr
After this you can: Generate and filter training data for open models.
Expert · 7 lessons · 17 hr 30 min
After this you can: Study how agent harnesses can improve themselves.
Your progress is saved in this browser only. Nothing is sent anywhere.