Start here — choose your track before Day 1
Everything in this programme is hands-on. Before you write a line of code you need somewhere to run it, and there are two ways to get that. Pick one now. Same curriculum, same assignments, same certificate — the only difference is where your code runs.
The single question that decides it: can you put a payment card on an account? If yes, take Track A. If no — or you would rather not — take Track B, and lose nothing that is examined.
WHEN YOU ENROLChoose your track on the enrolment form. You can change it later.
SESSION 1Cloud concepts. Nothing to install, nothing to set up. Just read and answer.
SESSION 2 — SETUPWalks you through your chosen track, click by click. The only session where setup happens.
FROM SESSION 3You are building. If Session 2 did not work, say so that day — do not push on quietly.
Enrol and choose my track →
The two tracks in detail
TRACK A
Your own Azure account
For anyone who can put a card on an account.
- What you need before Day 2
- An email address you control
- A payment card — for identity verification only
- A phone that can receive an SMS code
- About 20 minutes
- What it costs
- Nothing. The Azure free tier gives you $200 of credit for 30 days plus services that stay free. This course uses a small fraction of that. Microsoft will not charge the card unless you explicitly upgrade.
- What you will set up on Day 2
- A free Azure account
- A resource group of your own
- A Microsoft Foundry project inside it
- Your first model deployment, with its endpoint and key
- Why choose it
- You finish with genuine hands-on Azure experience — the real portal, real resource groups, real role-based access control. That is a line on your CV that Track B cannot give you.
TRACK B
No card required
For anyone who cannot get a card, or would rather not give one.
- What you need before Day 2
- A Google account — free, no card, takes two minutes
- Nothing else. No card, no phone verification, no billing details.
- What it costs
- Nothing, genuinely and permanently. There is no trial to expire and no credit to run out.
- What you will set up on Day 2
- A free Google AI Studio API key
- Access to our simulated Foundry portal
- Your Colab notebook, wired to a live model endpoint
- Why choose it
- You write exactly the same Python against a real, live AI service, and you learn the Foundry portal through a simulator built to mirror it. Every assignment, every assessment and the certificate are identical.
Said plainly: Track B teaches you everything the AI-901 exam tests, because the exam is a written paper about concepts. What it cannot give you is time inside the real Azure portal. If you can get a card, Track A is the better choice. If you cannot, Track B is a real path and not a lesser one — take it and start.
Side by side
| | Track A | Track B |
|---|
| Payment card | Required (not charged) | Not required |
| Cost to you | £0 / ₦0 on the free tier | £0 / ₦0, permanently |
| Where your code runs | Microsoft Foundry on Azure | Google AI Studio endpoint |
| Portal experience | The real Azure portal | Our simulator, built to mirror it |
| Setup time on Day 2 | About 20 minutes | About 10 minutes |
| Assignments | Identical — all 25, same rubrics |
| Assessments and final | Identical — same papers, same pass mark |
| Certificate | Identical — the certificate does not name your track |
| Can you switch later? | Yes, at any point. Tell us and we will move you. |
Plans
Every plan carries all 25 sessions and the same content. What changes is how much of it is marked, and whether it ends in a credential.
SELF-STUDY
$49
- All 25 sessions
- All 7 assessments, marked automatically
- Track A or B
- No written feedback
- No certificate
GRADED — MOST CHOOSE THIS
$129
- Everything in Self-Study
- All 25 assignments marked individually, each returned with a written letter
- The 60-question final, ten questions of which are built from your own mistakes
- Certificate, reference letter and programme outline on completion
MENTORED
$299
- Everything in Graded
- Two live one-to-one reviews with R. Wingo
- Limited places, because they are my hours
Have a promo code? Enter it on the enrolment page and it will set your plan automatically. Codes are also how paid plans are activated — you receive one when you buy.
How this works
1 · LearnOne module a day, thirty minutes. Open the week, work the session.
2 · DoEvery session ends in an assignment you build and submit.
3 · Get markedGraded against a published rubric. A written letter comes back, usually within minutes.
4 · FinishAll 25 passed, then a 60-question final built partly from your own mistakes.
What happens when you submit
No mystery, and no waiting days for a mark. Here is the whole loop.
STEP 1You submitEvery session ends with a form. Answer it in your own words and send. Nothing to email, nothing to attach.
STEP 2It is markedAgainst that session’s published rubric, criterion by criterion. The same rubric every time, for everybody.
STEP 3A letter comes backUsually within minutes. Not a score — a letter naming what you got right, what you got wrong, and what to do next.
STEP 4It lands on your timelineYour score joins your progress chart and the letter is saved to your account, so you can re-read it before the exam.
A real marked assignment
This is genuinely what comes back. Session 21, marked against its four criteria.
AI-901 — Session 21 — Graded
Content Understanding: Documents and Forms
94 / 100
The rubric, criterion by criterion
| Criterion | Awarded |
|---|
| Ran a real document through a prebuilt analyzer, with evidence | 25 / 25 |
| Three fields, values and confidence scores; identified the weakest and why | 30 / 30 |
| Structured vs unstructured data; when custom beats prebuilt | 21 / 25 |
| A real local application of document extraction | 18 / 20 |
What you did well
- You ran the experiment nobody asked for. Re-submitting the same receipt and noticing the confidence scores moved between runs is independent scientific behaviour, and it is the best thing on the page.
- Your least-confidence diagnosis is genuinely sophisticated — you identified competing monetary figures as the cause, not image quality. Most people at this stage say “the picture was blurry”.
- Every figure you quote matches your screenshot. No inflation, no fabrication.
What to fix
- Every confidence score came back low — 66.7%, 55.0%, 53.6% — and you reported that without asking why the whole extraction was weak. Look at which analyzer you pointed at which document, then re-run it against the other one and compare. Ten minutes.
- You wrote the answer to Part 3 inside Part 4 and did not notice. Join them next time — connecting your own answers to each other is how an argument becomes an analysis.
Next step
Re-run the same document through the other analyzer and send me both sets of confidence scores side by side.
Nothing below 70 is ever sent as a bare grade. If your work does not clear the pass mark, it comes back
as a letter telling you what is missing and what to do about it, reviewed by a human before it reaches you.
You resubmit. There is no limit on resubmissions and no penalty for needing one —
the point is that you end up understanding it, not that you got it right first time.
House rules
- Type the code yourself. Copying and pasting is not permitted. Typing is how it sticks, and the final will find out whether it did.
- Never share an API key. Not in a screenshot, not in a form, not with a classmate. A key authorises real spending on a real account.
- Report a blocker the same day. Send the exact error text. Every developer alive gets stuck; the good ones just ask sooner.
- The pass mark is 70. Below that, work comes back with what to fix rather than a grade.
- This is an ERW CommandPMO credential, not a Microsoft certification. It prepares you for Microsoft Exam AI-901, which you sit with Microsoft in your own name.
