Last week, I took both AWS Machine Learning certifications: the Machine Learning Engineer – Associate & Machine Learning – Specialty. I have to say, it was a long and exhausting journey, I can’t remember the last time I had to study and learn so many new things in such a short period of time, but in the end it was worth it, and I managed to pass both tests.
The Machine Learning Engineer Associate

First I did was the associate one, in the last day of the beta exam. I took the exam online, at home, and I finished it after around 95 minutes. With a total of 85 questions, the score needed to pass the exam was 720, and I scored 736 (yes, only 16 points over the limit 😅)🏅
The exam was way harder than I expected as it goes really deep on SageMaker algorithms, hyerparameters, ML techniques and integrations but also includes resources like Redshift ML, AWS Glue (DataBrew mainly) and EMR. Honestly, this is the level of exam I expected for Specialty, not Associate.
The Machine Learning Specialty

Trying to reuse some of the time I had already invested doing the ML Associate, I decided to do the Specialty straight afterwards, one week later, so I took the exam also online, at home, and I finished it after around 65 minutes. With a total of 65 questions, the score needed to pass the exam was 750, and I scored 810🏅
After doing the Associate one, the Specialty felt like a walk in the park, as I already knew what to expect and at what level. The scope was smaller, as most of the AI resources we have today did not exist when this exam was created, but on the other hand, the exam still includes some discontinued resources (e.g., Amazon Lookout).
ML Engineer Associate vs ML Specialty
The difference in the age of the tests is clear in the format of the questions and answers, but also in the resources presented. Based on my experience, this is how the exams differ in the main topics:
Topic | ML Engineer Associate | ML Specialty |
AWS Sage Maker | • What features to use (DataBrew, Canvas, Monitor, Registry, etc) • What algorithm to use for a specific issue | • What algorithm to use for a customer scenario • How to improve the algorithm performance • How to integrate with other tools/resources |
ML techniques | • How to train data • How to measure ML model efficiency | • Advanced techniques for pre-processing data • How to avoid bias • How to improve performance |
Other AI resources | • Redshift ML • AWS Glue • Amazon Bedrock • Amazon Comprehend • Amazon Transcribe • Amazon Kendra • Amazon Q • EC2 for AI (Inferentia and Trainium) | • Kineses DataStream, Firehose and Analytics • EMR • Amazon Kendra • Amazon Lookout • Amazon Comprehend • OpenSearch |
How I studied
For the ML Engineer Associate, we still have little content available on the internet, so I used the Stephane Maarek and Frank Kane course available on Udemy.
For the ML Specialty, most of the topics were also covered in the ML Engineer course, so to complement it, I did a few practice exams on Whizlabs.
Any questions? Let me know 🙂
Discover more from contains(cloud)
Subscribe to get the latest posts sent to your email.
Be First to Comment