I am learning this in public, and teaching it as I go.
I did not come from tech. No degree in it, no job in it, no team behind me. What I have is a run of real systems I built and now maintain, and a very clear memory of not understanding any of it.
Why this course exists
Every AI course I found was one of two things. Either it taught you to chat with a model and called that a skill, or it assumed you already knew what an API was and skipped the part where you get stuck.
So it is written for the person I was: a capable adult who can follow a real explanation, has no patience for filler, and wants to end the hour with something that works rather than a feeling of progress.
That is why every exercise leaves a file behind, and why the last track is about getting paid. Learning this is only interesting if it changes what you can do on Monday.
- A multi-tenant learning platform — this one — with 38 tables, 73 row-level security policies and an automated suite that drives the live site before anything ships
- A live conversational agent handling real customer enquiries, with its rules in code and around 250 tests holding them in place
- A nine-provider model adapter behind one interface, with streaming and per-request cost accounting
All of it learned the way this course teaches it.
What I will not do
Pretend it was easy
The videos include the parts where it breaks and I have to work out why. That is the actual skill and cutting it out would be a lie.
Quote a number I cannot source
If a company does not publish a price or a limit, the lesson says so and teaches you to check it yourself.
Waste your tokens
No exercise exists to be cute. If a lesson spends money, the spending is the lesson.