I recently took my first AWS certification exam, and went straight for ML Specialty Certification, as I felt that certification was the most relevant to my situation.
The exam will test you to answer 65 multiple answer/multiple choice questions in 180 minutes in four areas:
- Data Engineering
- Exploratory Data Analysis
- Modeling
- ML Implementations and Operations
The test does not ask any python or R code specific questions, but will test you on understanding of AWS and modeling concepts. Read the details of exam here.
I am a trained physicist turned machine learning engineer with about 4.5 years of experience with machine learning methods and the AWS services.
In my current company, we use AWS services and deploy machine learning models, so I have had experience in some of those topics, but have never used services like SageMaker before (we manage our own Kubernetes cluster on AWS).
For me, it took me about 1.5 months to prepare for this exam. When taking the sample questions on AWS website for this certification, I realized exam questions were very detailed (which was also the case in the real exam). So taking the test without preparation is NOT recommended.
Also, if you have less than two years of experience with AWS, it is recommended to take other certifications first. If you are new to machine learning, I would recommend not taking this exam until you have experience putting a couple of models into production.
Fortunately, there are a lot of materials available online to prepare you for this exam, and I used some of these materials to be able to pass the exam on the first try.
- Udemy course: This is the main course I took to prepare for the exam. The course is very clear and organizes subjects in a logical manner. The instructors keep the course updated with new materials as AWS roles out new ML capabilities.
- AWS exam readiness: AWS offers its own learning path and exam readiness course, I did not take as agenda seemed very close to the Udemy course I took.
- AWS documentation and FAQs on various topics: I would find if reading all AWS documentation, all answers to the exam questions can be found. However, reading all of those documentation is a time consuming task. I referenced to these documentations once I needed to learn more depth on a specific topic.
- Cloud Guru course: I did not take this course, but heard it has great context as well.
- My previous people manager also had very extensive notes on his github repository for taking the AWS ML specialty exam which I found useful for comparing notes.
Once feeling comfortable with the materials of the course, I recommend taking as many practice exams as you can. Practice exam will pressure test you and some questions will challenge you. As you take more practice tests, you challenge your knowledge, and may learn more concepts to track on your notes for future practice tests.
I took 10 practice tests, my first practice tests were around 70%, but reviewing the materials, over the next practice exams it increased to about 90% (I found it difficult to get the full mark in any exam, as some questions will challenge your depth of knowledge in various areas). In my actual exam I got a scaled score of 877 (minimum passing score is 750). Comparing notes with others and the forecasting, led me to think your expectations from the practice exam can be extended to your actual exam. I think it is important that you feel good about your practice exams before taking the actual exam.
There are so many resources for taking practice exams, and took some of them:
- AWS sample questions and official free practice test. Use the result of these tests as a good proxy for how much preparation you need to take the actual exam. Some of the questions will challenge you (like the actual exam).
- AWS official practice exam: Very much like the actual exam, will give you confidence if you want to be tested very similarly to the actual exam.
- Udemy practice exam 1 and practice exam 2: I found two exams on Udemy which I took as a follow up to the course I took.
- Cloud Guru practice exams: A good source of questions you can test yourself with. Did not take this exam.
- Test Prep practice exams: I took a series of practice exam tests on this website too.
- Whizlabs practice exams: Another good source practice exam you can test yourself with. Did not take this practice exam.
As you take practice exams, you find some questions are repeated, which is also the case for the actual exam. So take practice exams often. In each exam I took, I found about 10% new questions which I have never seen before, which was also the case for the exam. So you can use practice sets to increase your knowledge on the areas you need improvement. It is important to learn from mistakes you made in the practice exam, and read detailed AWS documentation on the subject.
One thing I have to say, you can never over-prepare for this certification exam, so study often and in depth. But will schedule the exam once you feel confident enough.
Looking back, taking the exam helped me organize my thoughts about how machine learning in the AWS ecosystem works.
Here, I put the notes I took throughout my study for the exam. Hope you find this useful. I highly encourage you to take your own notes as well, and compare notes with your colleagues.
Never stop learning!