My path to AWS AI Practitioner certification: An experience report

The economic situation is not exactly making life as an IT freelancer easy at the moment. At the same time, the demand for AI experts is likely to rise sharply in the near future. As an AWS Cloud Architect with several AWS certifications already under my belt, specializing in AI seemed like the logical next step.

The topic of AI is complex, but with a solid IT background, you can get started relatively quickly through targeted further training. The AWS AI Practitioner certification was the perfect entry point – also because it can serve as preparation for the more demanding AWS Certified ML Associate exam.

Incidentally, I first came into contact with AI back in the 1990s during my studies at TU Berlin, when I attended a seminar on neural networks and genetic algorithms. Since October last year, I have been working more intensively on the subject again.

I invested a total of 3-4 weeks in preparation. I used the Udemy course “Ultimate AWS Certified AI Practitioner AIF-C01” by Stephane Maarek as my main resource. The course teaches the necessary basics to pass the exam, but does not go into depth on many topics. Especially the architecture of AI models and the underlying algorithms are only covered superficially – it’s not enough to really understand how LLMs work.

After about 60% of the course, I already started taking mock exams. I used the Udemy course “[Practice Exams] AWS Certified AI Practitioner – AIF-C01” by Stephane Maarek and Abhishek Singh. It contains four mock exams with 65 questions each, which are very similar to the real exam. I narrowly passed the first mock exam.

My learning strategy was to work intensively on the questions I answered incorrectly. The answers often contain links to AWS documentation, blog articles and white papers. Working through these sources proved to be an efficient method to close specific gaps.

For me, the biggest challenges were the questions on machine learning models and algorithms. Many concepts are asked here that are only touched on in the course, without deepening them with practical examples.

To master these and other difficult topics, I studied with Anki flashcards. In total, I created around 100 cards – in comparison, I had created and studied around 2000 Anki cards for the “AWS Certified Solution Architect Professional” exam. The effort required for the AI Practitioner certification was therefore significantly less.

Incidentally, I found the topic “AWS Bedrock” the most interesting because it offers an application-oriented introduction to the topic of AI. This service makes it relatively easy and inexpensive to develop and operate AI applications – in contrast to the complex and computationally intensive training of your own AI models with SageMaker.

The exam itself was almost routine for me, as I had already completed four AWS certifications last year. I decided to take the exam from home and unfortunately had technical problems with the system check again. The OnVue app was not stable and kept crashing my MacBook Pro. The evening before the exam, I analyzed the problem late into the night and finally found a solution: I set up a new user account on the Mac and only installed the OnVue app.

The actual exam then went smoothly. As a non-native speaker, I was given a 30-minute extension so that I had enough time for the 65 questions. There were no major surprises during the exam. Of course, there are always some questions where I was in doubt or had to guess, but the mock exams had prepared me well and boosted my confidence.

Compared to my previous AWS certifications, I found this exam relatively easy. Long questions with long answer options are generally more challenging as the complexity often increases with the length – but there were only a few of these in this exam.

If I were to go through the exam preparation again, I wouldn’t change anything fundamentally. I think I found a good balance between learning effort and exam success.

Ideally, I would have invested more time in hands-on tasks to better understand the topics. But this is time-consuming and not necessarily efficient in terms of exam success. I’m also not a big fan of scripted tutorials. With 30 years of professional experience in IT, I have enough experience to work through the topics myself.

With today’s information overload, focusing on relevant topics is one of the most important skills of an IT professional. The scope and complexity of IT technology in general and AI technology in particular is almost infinite.

The AWS AI Practitioner certification only provides a rough overview of the topic of AI and the offerings from AWS. For me, it mainly serves as preparation for the AWS Certified ML Associate certification, which offers much more depth and represents real value in one’s own knowledge portfolio.

While you do get a good overview of the basics of AI and the AWS services during the preparation, there are other ways to do this that save the hassle of an exam.

The course helped me to better define my own path in AI technology. I learned about AWS services that I can use to create my own offering – especially in the area of programming AI applications such as LLM Chains and Retrieval Chains that can access Bedrock services. I also gained an insight into the complexity and cost of training AI models – an area that is often only affordable for large organizations and requires a much deeper dive into machine learning.

My next step will be the AWS Certified ML Associate exam. This exam is much more demanding and requires a deep understanding of AI models and algorithms. In my estimation, the “AWS AI Practitioner” exam covers about 30% of the material of the “AWS Certified ML Associate” and is therefore a good intermediate step.

Courses:

• Ultimate AWS Certified AI Practitioner AIF-C01 by Stephane Maarek
• Practice Exams AWS Certified AI Practitioner – AIF-C01 by Stephane Maarek and Abhishek Singh

Learning methods:

• Anki – powerful, intelligent flashcards: Anki app for flashcard learning
• Image Occlusion Enhanced – AnkiWeb: Plugin for efficient flashcard creation from course slides

Exam execution:

• For Mac problems with OnVue: Try a separate user account for the exam app only
• Use the 30-minute extension for non-native speakers

AWS services to go deeper:

• AWS Bedrock for practical AI applications
• AWS SageMaker for deeper ML understanding