Back to index
UdemyAbout 25 hours of video across 14 sections and 318 lectures, with 110 quiz questions and a 65-question practice exam·Udemy sale pricing, usually around 15 dollars or euros, or included in Udemy Personal Plan. 30-day refund. Optional paid hands-on labs. The AWS exam is a separate 150 dollars

AWS Certified Machine Learning Engineer Associate: Hands On! (Frank Kane and Stephane Maarek, Udemy)

4.3

The most complete video course for AWS's ML engineer exam, and already updated for the September 2026 MLA-C02 rewrite. Kane brings real ML knowledge and Maarek brings the AWS plumbing, but it is long and a bit uneven in places.

What We Liked

  • Updated for MLA-C02, the revised exam that now covers generative AI, agents and foundation models
  • Frank Kane's years at Amazon show in the ML sections, which explain the why and not only the service names
  • Strong on data ingestion, feature engineering and SageMaker built-in algorithms, the areas people usually fail on
  • A full 65-question practice exam plus 110 quiz questions throughout
  • Cheap relative to the exam fee if you catch a Udemy sale

What Could Be Better

  • 25 hours is a lot of watching, and some sections are more slides than demos
  • Two instructors with different styles means the course does not always flow
  • Proper hands-on practice costs extra, and SageMaker demos can run up a real AWS bill if you follow along carelessly
  • One practice exam is not enough for an associate-level test, so budget for a question bank

Detailed review

This course pairs two of Udemy's biggest names in this space. Frank Kane spent about nine years at Amazon and IMDb and runs Sundog Education, and Stephane Maarek is the go-to AWS certification instructor. Kane covers most of the machine learning, Maarek covers a lot of the AWS infrastructure, security and operations side. When I checked, the listing showed about 65,000 learners and a 4.6 rating from over 6,000 reviews, with an update this month.

Timing is important here. AWS replaced the MLA-C01 exam with MLA-C02, opening the English beta in September 2026 and retiring the old English version at the end of that month. The revision widens the role to include generative AI, agentic AI and foundation models, so a lot of older prep material is now stale. This course has already been rewritten for C02, and you can see it in the syllabus, which now has sections on generative AI model fundamentals, building applications with Bedrock, and MLOps and what it calls AIOps, alongside the classic data ingestion, transformation and feature engineering, managed AI services, SageMaker built-in algorithms, and model training, tuning and evaluation.

At roughly 25 hours across 318 lectures, it is long. I would not try to watch it start to finish. Skim what you already know, and spend real time on the data preparation and SageMaker training sections, because in my experience those are the parts of this exam that trip up software engineers. Kane is good at explaining things like class imbalance, regularisation and hyperparameter tuning in plain language before showing the AWS way of doing them, which makes this more educational than most cert courses.

Maarek's sections on IAM, VPCs, encryption and monitoring are dry but necessary. The weaknesses are what you would expect. Switching between two instructors is a little jarring. Some sections lean on slides more than live demos, and if you want guided labs Udemy charges extra.

I would rather spin up a sandbox AWS account with a budget alarm and repeat the key SageMaker and Bedrock steps myself, which costs a few dollars if you clean up properly. You also only get one 65-question practice exam, which is not enough for an associate exam. Add Tutorials Dojo or the official AWS practice set. Overall, if you are going for this certification right now, this is the course I would buy, mainly because it is current and because Kane actually teaches the ML rather than just naming services.

[ final ]

The verdict.

Best single course for the AWS ML Engineer Associate, especially now the exam has changed and many older materials are out of date. Use it if you already work with AWS or ML and want structure. Add a set of practice exams, and do at least some of the SageMaker and Bedrock work yourself in a sandbox account rather than only watching.