12 modules · ~81 min
An AI Project, End to End
One company, one problem, twelve desks. Follow a single AI project from the boardroom argument about a revenue decline all the way to the meeting where someone proves whether it worked — and see exactly what each person decides, and what they hand to the next.
This course is free — no account needed. Dive in.
Finish the course, earn a verifiable certificate. You get a unique credential ID and a public verification page — and you can add it to your LinkedIn profile in one click.
What you'll be able to do
Follow one real problem through every person who touches it, in order
Understand what each of the ten roles actually decides — and why
See how a decision made in the boardroom quietly sets the limits for the data scientist five months later
Read a hand-off: what lands on a desk, what gets added, what moves on
Explain why an AI project succeeds or fails for reasons that have nothing to do with the model
Recognise the four places these hand-offs usually break, and what breaking looks like
Speak confidently with strategy, finance, product, data, engineering, governance and operations colleagues
Judge a finished AI project the way a board does: impact, cost, evidence, risk
Modules
12 in this track- 1How One Problem TravelsFreeA 15% revenue drop lands in a Monday meeting. Ten people will touch it before anyone knows whether it was fixed — and each one hands something to the next.Start →
- 2The CIO: Where the Money GoesFreeDaniel has $600k, seven teams asking for $1.1m, and no idea why customers are leaving — so he funds the two things that can answer that question and tells marketing no.Start →
- 3Finance: Is It Worth It?FreeOmar turns Daniel's mandate into two numbers with teeth — a $250k spending envelope and the smallest churn reduction that would still have been worth paying for.Start →
- 4Product: What Exactly Are We Building?FreeSofia turns a mandate to cut churn into a requirements document — sorting goals from features, drawing the MVP line, and choosing the four warning signals the model will be allowed to learn from.Start →
- 5Customer Success: What Is Actually HappeningFreeA dashboard flags 120 accounts as low engagement. Maya has six people and knows most of those flags are wrong, so she sits down and labels what is really going on — and those labels become the truth a model will learn from.Start →
- 6The Analyst: Where and When Did This Start?FreeEveryone has been arguing from anecdotes. Ben has the tables — and by Friday he can say which month the decline began, which customers it is eating, and that leavers go quiet ten weeks before they cancel.Start →
- 7The Data Engineer: Making It Arrive Every NightFreeBen's answer took three weeks and forty hand-run queries. Aisha has to make the same thing happen unattended at 2am, every night, and keep being right when a source system quietly renames a column.Start →
- 8The Data Scientist: Who Is At Risk?FreePriya trains the model everyone has been waiting for — then throws away her best feature because it was cheating, and sets the alert threshold from how many phone calls six people can make in a week.Start →
- 9Governance: The Conditions Under Which Yes Stays SafeFreePriya's model works, and Sales has already asked for it. Elena spends one week deciding what the score is allowed to do, who may see it, how long it lives and who owns it when it drifts — and hands Kofi five conditions rather than a signature.Start →
- 10The ML Engineer: From Notebook to Monday MorningFreePriya's model runs on her laptop, Elena's conditions are taped to the monitor, and Sofia has three small additions. Kofi has twenty story points to turn a prototype into something that survives a bad Tuesday.Start →
- 11Operations: What We Actually Do About ItFreeSixteen alerts land on Jade's screen on a Monday morning. She has six people and no spare hours, so she matches each response to the cause behind the score — and deliberately does nothing about three of the eight accounts in front of her.Start →
- 12Did It Work — And What Happens Now?FreeNadia leaves a quarter of the at-risk accounts alone on purpose, refuses to call the result early, and hands Daniel a number small enough to be true — then the board decides what to scale, what to fix, and what to stop.Start →