After 4 months of intense studying I passed the Google Cloud certification for Professional Machine Learning Engineer!
Google Cloud Platform certificates are considered to be challenging compared to other cloud platforms .
What it means to be a certified machine learning engineer for Google Cloud?
The certificate is a proof of skills for Google Cloud Platform machine learning products. Additionally the certificate exam covers wide range of topics from machine learning concepts, Tensorflow and MLOps.
Here is the full list of machine learning engineer exam requirements:
- Frame ML problems
- Architect ML solutions
- Design data preparation and processing systems
- Develop ML models
- Automate & orchestrate ML pipelines
- Monitor, optimize, and maintain ML solutions
The Professional Machine Learning certificate is at the most challenging level among the Google Cloud certificates:
GCP exam is a bid deal!
When revealing my certificate attempts, I told I would not stress about it.
So ofcourse I stressed a lot.
My continuation in the project depended on the completion. Also, I like Google Cloud over the alternative cloud platforms and machine learning is definitely the field I want to be in. This was a great chance to prove my skills in advanced areas such as neural networks. I really wanted to pass.
The exam itself costs 250 $.
No feedback is given, it is only pass or fail. I you fail, you can not be sure how to improve.
In the exam you have barely time to skim trough the verbose questions and answer alternatives sometimes extending to multiple lines.
I got a bit over 70 % correct in two practice exams. Some sources estimate that even 80 % is required to pass the actual one.
How would you not be stressed in such situation?
My learning stats before GCP MLE certification
|GCP hobby experience||1.5 years|
|GCP work experience||5 months|
|Exam preparation time||4 months|
|Exam type questions and answers reviewed||235|
|Videos watched||15 hours|
|Blog posts written||16 posts|
|Google Cloud docs||Product docs, architecture best practices, ML fundamentals|
|Total hours spent||Definitely more than 25 h estimated by the customer who required the certification…|
The GCP exam day
I was really nervous on the exam day. The work hours were not the most efficient. The exam was booked at 16:00.
15 minutes before the scheduled time I logged in to the Webassessor system that Google uses as their certificate exam platform. I closed Teams, Outlook and other software on my computer. The private browser for online exams was easy to download.
The Launch button became active 15:50 as promised. The instructor asked a few questions through chat. Needed to show around my cleaned bedroom that had was transformed to my exam chamber.
Showing an ID was mandatory. I had left phone to another room earlier and shut it down.
Time management in the GCP exam
Time matters when you have 60 challenging case questions in 120 minutes. So I had developed a personal question reading technique:
- Read the question and answer options super quickly
- Do not stop
- Do not try to understand details
- Identify the important part of the long question
- Read again slower to grasp the details
- Answer and move to next question
The first 30 questions took sligthly more than 30 minutes. The other 30 took around 40 minutes. Fatigue was clear towards the end.
35 minutes was spent to go through the 34 questions I had marked for review. To tag optimal number of questions for review, you should quickly estimate how confident you are about the answer.
After this I still had 15 minutes. I had three really tricky ones. After reading carefully they made a bit more sense.
The last 10 minutes was enough to go through the first 30 questions again. I just read the chosen option and estimated whether it still made sense. Actually I still changed 2 answers on this round.
Passing the exam
After the time was up I was exhausted. The result could be equally well be either pass or fail.
I assured myself that I had given everything. A few deep breaths and it was time to check the results. Pass!
The final verification of the results took 9 days. Probably Google evaluated from the web cam recording and the answers that I did not cheat during the exam.