learning · ai · 23 September 2026
Studying for AWS AI Practitioner with OpenMAIC
I used a self-hosted OpenMAIC classroom alongside Stéphane Maarek’s Udemy course and practice exams to prepare for AWS Certified AI Practitioner. On Sunday, I sat the proctored exam at home—and passed.
On Sunday at 1.30 pm, I sat the AWS Certified AI Practitioner exam at home with remote proctoring. I passed.
My preparation combined a really good Udemy course by Stéphane Maarek, practice exams, and a slightly more experimental addition: a self-hosted AI classroom built with OpenMAIC . The certification is the result, but the classroom is a project worth explaining in its own right.
It also gave me a practical use for AI while studying AI: turning reference material into a structured course I could revisit, with lesson pages, quizzes and narration in one place.
What is OpenMAIC?
OpenMAIC stands for Open Multi-Agent Interactive Classroom . It is an open-source project for creating interactive learning experiences with AI. You can describe a subject or supply learning material, then generate and revise a course rather than work through an endless chat transcript.
The project includes AI teacher and peer interactions, slides, quizzes and interactive activities. Its Pro workbench adds a course-building agent that can plan a curriculum and edit individual pages. It also supports different model providers and persistent course storage. Those are project capabilities; which ones are available in a particular installation depends on how it is configured.
My use was focused: build a reusable AWS AI Practitioner revision course and use it alongside established teaching material.
The course behind the screenshot
The screenshot shows my 32-page AWS AI Practitioner course in the OpenMAIC workbench. On the left are the lesson pages. The main canvas shows the roadmap, with narration actions underneath.
The course is organised around a nine-day rhythm:
Start with a diagnostic and AI/ML foundations.
Move into generative AI foundations and foundation-model applications.
Cover responsible AI, security and governance.
Revisit weak topics, then work through mixed practice and a final light review.
That is the structure of the course, rather than a claim that the whole qualification can be learned in nine days. The roadmap suggests 25–40 minute blocks, teaching followed by a quiz, and review pages for missed questions.
The distinction between covering a topic and being able to answer a question about it matters. A page on precision and recall can look familiar; a question that asks which metric matters in a particular situation makes you apply it. Putting explanations and checks next to each other makes that distinction visible.
One useful reminder is written directly onto the roadmap: practice accuracy is not the AWS scaled exam score . A percentage from a generated quiz should not be treated as a prediction of the final result.
The Udemy course and practice exams still mattered
The main taught course I used was [NEW] Ultimate AWS Certified AI Practitioner AIF-C01 by Stéphane Maarek . I thought it was really good. I accessed it through Udemy Business, and I also did practice exams as part of my preparation.
OpenMAIC was an addition to that preparation. The roles were different: the course provided the teaching, practice exams provided another way to check understanding, and OpenMAIC held the custom revision material in a format I could work through again.
There is a useful boundary here. A generated classroom is not automatically an authoritative exam bank. It can organise material, explain it in another form and generate questions, but the official exam scope and reliable source material still need to anchor it.
For this project, the course materials were supplemented with the official AWS exam guide. That matters because both cloud services and exam coverage change; a polished-looking lesson is not proof that it covers everything currently required.
Running my own study workbench
I host my installation privately, with persistent storage for the course, uploaded material and learning state. The service has its own database and backups, so it is intended to be a reusable study workbench rather than a temporary browser session for one exam.
The installation also supports saved lesson narration. Generated audio can be stored and replayed instead of being regenerated each time a lesson is opened.
There is an important practical distinction in the setup: self-hosting the application does not mean all the AI runs locally . My installation uses a cloud model provider. Material supplied to those model calls leaves the local application, and generation has usage costs. Keeping the workbench private controls access to the service; it does not turn cloud inference into an offline process.
I am sharing the workbench screenshot here, not redistributing the underlying paid course material.
AI-generated study material needs checking
The classroom needed review before it was useful. During setup, some generated questions were off-topic, some narration needed correcting, and parts of the service explanations needed adjustment. The lesson layouts also needed work to keep text readable.
That is a fairly grounded picture of this kind of tool: it can create a substantial starting point, but the starting point still needs checking. A confident explanation and a tidy slide can both be wrong.
For anyone trying a similar approach, the practical sequence is straightforward: start with dependable teaching material, check the current exam guide, generate a focused revision course, and inspect its explanations and questions. Keep the practice exams in the mix. Use mistakes to decide what needs another pass rather than treating completion of a slide deck as readiness for the exam.
The result
I took the remotely proctored exam from home on Sunday, 20 September, at 1.30 pm and passed AWS Certified AI Practitioner .
That is the outcome of the combined preparation. It does not establish how much of the result came from any one resource. What it does leave me with is both a certification and a working OpenMAIC study environment that can be adapted for another subject.
My verified credential is below, with a direct link to the public record on Credly.
View my AWS Certified AI Practitioner credential on Credly .