The Azure AI-900 is the first step in mastering AI Engineering on the Microsoft stack. It’s an introduction-level exam, so there’s no expectation of previous experience with AI or ML. I recently passed my Microsoft AI-900 exam, and here’s how I did it.
Main Topics of Study
The AI-900 exam is mainly a test of basic AI/ML concepts and the Azure services you can use to use those concepts. For example, if you want to read the text of a stack of paper documents, you would use Optical Character Recognition (OCR). In Azure, you can use the Vision service to perform an OCR workload. Most of the questions are multiple-choice, and you won’t have to write code or read case studies. There are many questions where you match a particular workload to the correct machine learning technique or Azure Service. If you understand the various types of machine learning and AI workloads and how to run them on Azure, you will be golden.
Describe Artificial Intelligence workloads and considerations
This objective covers the different types of AI services on Azure and how to use them. You need to know the difference between Azure AI Services (aka. Cognitive Services), Azure Machine Learning, and Azure Bot Services.
Describe fundamental principles of machine learning on Azure
This objective covers general machine learning techniques. To succeed here, you must understand the basic classes of machine learning algorithms and when to use them. This includes topics like regression, clustering, and classification. You also need a high-level understanding of Azure Machine Learning Studio.
Describe features of computer vision workloads on Azure
This objective covers the variety of services under the computer vision umbrella. This includes services like Computer Vision, Face, and Form Recognizer. You’ll also need to understand the basic concepts of Computer Vision, like image classification, OCR, and object detection.
Describe features of Natural Language Processing (NLP) workloads on Azure
You’ll need to understand the various NLP features on the Azure Stack, primarily focused on the Language service. You must understand the various features of translation, sentiment analysis, and text analytics. You’ll also need to understand the concepts of NLP, like tokenization, lemmatization, vectorization, and stemming.
How to Prepare for the AI-900
The AI-900 exam page has a collection of self-paced MS Learn training to prepare for the exam. This is your primary resource for the test. Don’t bother with buying a book or hunting around for a course. Just do the free training.
After completing the training, do the free practice exam. If you can consistently score over 80%, you’re ready to take the exam. If not, you’ll get a custom list of MS Learn Modules to complete, and then you can try again. The practice exams are free, so keep taking them. After completing the course, I took the exam every day until the test day.
I’d also recommend looking over the product descriptions in the Azure Docs. You don’t need to memorize them, but you should know what each service does. This will help you fill in any knowledge gaps from the MS Learn training.
Finally, I recommend watching this exam overview video on your test day or before doing the coursework. It’s a solid review of what the test is about.
Overall, this is not a challenging exam. I’m relatively new to the AI space, and it took me a couple of weeks to plow through the MS Learns and pass the exam. You also don’t need to pony up any dollars for training. The freely available training is more than adequate to ace this exam.
Should You Take It
I’m on the fence about developers taking certification exams. They don’t often translate to career capital, but the certification process helps measure mastery of a new topic area. Certifications give you an idea of what skills people think are valuable within a specific area. They’re worth the energy if you’re branching out into something new and want to ramp up your skills.
If you’re already an AI Expert or have no interest in Azure AI Services, I’d skip this one and spend time elsewhere. If you want to explore AI and ML and learn the Microsoft stack, hit the books, study, and add this credential to your resume.