[alfalfasprout]

As someone in this space I could not disagree more.

Certifications will do nothing for you. The harsh reality is only real world experience doing this stuff at scale will help you understand all the complexity involved. There are tons of people trying to hop onto this train after taking a few online courses and it's making it hard to filter down candidate pools.

[visarga]

I ask my job candidates simple/foundational questions and even those are hard for most of them. I don't care about the degree, but I care the candidate can access and reason with the core concepts. And I value programming skills a lot.

[burkaman]

As someone outside of this space who has taken a few of the well-regarded ML and deep learning Courseras, I agree with you. You can get a certificate without learning a single thing and just putting in a couple hours of work, and even in good faith it's hard to get a ton out of it since the assignments are so shallow.

I do think they can be valuable if they help you learn the basics and get started on a bigger personal project, but not as something to put on your resume.

[joshvm]

I would agree on one aspect though - deploying a model at scale is much closer to SWE than it is to foundational ML research. At a high level, you're deploying a function which has some known compute requirements. It requires setting up infrastructure, monitoring/logging, API setup etc. This is the sort of thing that a good devops engineer could probably make a horizontal move to, because a lot of the practical experience is similar. I don't think you need a particularly deep knowledge of ML unless you're also expected to be involved in trying to track model performance that might require re-training. Leaving aside the really distributed systems that require multi-node, multi-GPU (but again, if you have HPC experience, that should transfer).

The problem is a lot of tutorials just show you how to make a Flask/Gradio website (maybe FastAPI) and call it a day. A lot of the experience here is the sort of in the trenches practical stuff that you can't cover in a MOOC (and it's expensive to experiment with GPU clusters). I suspect there are better non-ML courses people could take though.