Welcome back to our journey π«±π»βπ«²π½
This time, I took the AWS Certified AI Practitioner exam. I’ve decided to try both new AI certifications while it still in beta, to feel the difference from the regular one (and because it has a cool early adopter badge π).
It was really challenging because there is still very little content available on the internet, and even the practice tests I normally use to study were not available, so I had to took a different route.
The AWS AI Practitioner exam
I took the exam online, at home, and I finished it after around 55 minutes. With a total of 85 questions, the score needed to pass the exam was 700 and I scored 779 π
The questions were short and straightforward (mostly had 1 line only), but the key difference of this exam is that many of the questions were not “AWS-related” and focused on AI concepts, which makes the test more difficult and comprehensive.
In my opinion, you must be comfortable with the following topics for this exam:
AI Concepts:
- Common Machine Learning techniques (Supervised learning, Unsupervised learning, Reinforcement Learning, Reinforcement Learning with Human Feedback (RLHF))
- What is a Foundation Model (FM)
- Differences between LLM vs SLM
- Generative Adversarial Network (GAN)
- Data Augmentation
- Prompt Engineer (how to influence AI to adjust the tone according to the company’s characteristics)
- How to avoid Bias
AWS related:
- For Amazon Bedrock: Guardrails and Knowledge bases
- For SageMaker: Clarify, FeatureStore, DataWrangler and GroundTruth
- What AWS offers for AI running on EC2 (GPUs SKU’s, AWS Trainium, AWS Inferentia)
Other AWS AI resources, such as Amazon Comprehend, Amazon Kendra, Amazon Textract, and so on, were also on the test but in a smaller quantity.
It’s important to note that, as this is a Practitioner exam, you don’t need to have in-depth knowledge of all AWS resources, but you must know the main features and the use cases to which they apply.
What I expected to see in the exam, but didn’t:
- Overfitting or Underfitting
- How Hyperparameters impacts the model training (only got questions related to Temperature)
- How to evaluate performance of models (which metrics or formula to use)
- Specific questions about available models, for instance, Amazon titan, Stable Difusion, etc.
π¨ These topics didn’t appear in my exam, but they may appear in yours.
How I studied
As expected for a beta test, there were not much content available on the internet related to this specific exam, so my first step was to go through the official AWS free content, here are some of the content I read:
- Fundamentals of Machine Learning and Artificial Intelligence
- Exploring Artificial Intelligence Use Cases and Applications
- Essentials of Prompt Engineering
They have nice content if it’s your first contact with AI, but as I’ve already had prior experience with most of these concepts, I ended up not finishing these learning paths and went in search of something more specific.
During my journey, the only course + practice exam I was able to find were both by Stephane Maarek on Udemy:
- [NEW] Ultimate AWS Certified AI Practitioner AIF-C01
- [Practice Exams] AWS Certified AI Practitioner – AIF-C01
Luckily for me, the course was updated 3 days before my exam with new content that appeared in the beta and what I liked most about the course are the hands-on videos, which allowed me to create visual memories of the features and functionalities.
Any questions? Let me know π²
Discover more from contains(cloud)
Subscribe to get the latest posts sent to your email.
Be First to Comment