Everyone calls themselves an AI expert now. Almost nobody can show you a system they shipped.
This is the curriculum we’d hand someone who wants to close that gap. Thirteen modules that take you from “I know what a model is” to “I built the thing, it runs in production, and it answers to a revenue number.” It’s free, it’s self-directed, and it points at the best material we’ve found rather than reheating it.
Every module works the same way: the idea in plain English, why it matters commercially, one thing to build, and the resources worth your time.
Who it’s for
People who sell, market, or run revenue and want to build the AI, not just buy it. Founders, RevOps and demand-gen people, and engineers who want the go-to-market context that makes their systems worth paying for.
You don’t need a math background. You do need to be willing to build things.
What you’ll get out of it, and what you won’t
You’ll get practitioner-grade fluency. Enough to architect a system, write the code, make the cost and quality calls, and hold your own with a VP of Sales and a senior engineer in the same meeting.
You won’t get research-grade depth. No deriving backprop, no convergence proofs, no PhD math. The goal isn’t publishing papers. It’s shipping AI systems that answer to a number: pipeline, revenue, meetings booked.
Every module carries a depth target so you don’t over-invest where it doesn’t pay off.
- Conversational means you can explain it to a client and know when it applies.
- Working means you can build it and debug it with the docs open.
- Deep means you can design it from scratch and teach it. Those are the ones worth an extra week.
The two tracks
Track 1 is the machine learning floor. A few weeks, enough to be credible and not make rookie mistakes. It isn’t the job, so move fast. Track 2 is the real work, building AI for revenue, and it runs for months and then never really stops, because the field moves every month.
The full module list, with depth targets, time estimates, and your own progress, is laid out below.
A cadence that works
- Weeks 1 to 6: Track 1. Fast and conceptual. Don’t polish it to death.
- Months 2 to 4: Track 2, modules 6 through 10, in order. The core of it.
- Months 4 to 6: modules 11 through 13, going deeper on 12 with real work.
- After that: a few hours a week on primary sources, and one thing shipped every month.
The one rule
No tutorial hell. A module isn’t done when you’ve watched the videos. It’s done when the thing you built is live.
Watching teaches almost nothing. Building teaches everything. And the things you build become the portfolio that makes the expertise claim unnecessary, because you can just point at them.
Start with Module 1 below. Your progress saves in this browser as you go.