Passed the Google Cloud PMLE in ~2 months.
I just passed the new version of GCP Professional Machine Learning Engineer and thought I would share what worked for me. My main issue was that I did not have enough Machine Learning experience, so I spent a lot time on hands on projects and google.skills. Biggest tip: do a lot of labs or small projects and then spend most of your time on quizzes. Obviously use AI to understand when you go wrong.
| Item | Comments |
|---|---|
| Mona Book | Skip, its dense and the questions don't feel similar to the ones you get on the exam |
| Google Docs | Very useful for me. Also the service level docs were the best study resource |
| https://quiz-trail.web.app/ | Free and great... Generally updated. But had a few code level / TF questions which I never got. |
| https://github.com/AndyTheFactory/gcp-pmle-quiz | Free and great...! A bit out dated now but a very good and long collection of questions |
| https://skillcertpro.com/ & https://www.whizlabs.com/ | Paid and low priority. The questions in the above 2 were more realistic. Also a lot of the answers were wrong. |
| https://www.skills.google/paths/17 | Google ML skills. Very useful for me. Probably the best $ spent. |
A lot of quizzes will have low level TF, or code or some weird config flag name. The real exam had none of that. All questions were scenario based "given blah blah..pick one". And no multi choice.
Besides these I also built a few simple models and pipelines to really understand the main concepts. Be careful with this cause if you aren't careful, you will get a few billing shocks 😂 ..in my case it was an AutoML job that cost me $40ish.
Out of the 50 total questions in the exam, I was pretty confident I got the right one in about 40. Found some of the env or pick the right model for this on weird scenario a bit tricky. And there were 2-3 questions that went completely over my head. Did the exam in person based on other recommendations here.