A former colleague of mine (SW guy) took Andrew's Coursera course, downloaded some Kaggle sets, fiddled with them, and put his Jupyter notebooks online. He learned the lingo of deep learning (no experience in them, though). Then he hit the interview circuit.
Got a senior ML position in a well known Fortune 500 company. Senior enough that he sets his goals - no one gives him work to do. He just goes around asking for data and does analyses. When he left our team he told me "Now that I have this opportunity, I can actually really learn ML instead of faking it."
If you think that's bad, you should hear the stories he tells at that company. Since senior leadership knows nothing about ML practices, practices are sloppy to get impressive numbers. Things like reporting quality based on performance on training data. And when going from a 3% to a 6% prediction success rate, they boast about "doubling the performance".
He eventually left for another company because it was harder to compete against bigger charlatans than he was.
If he really did take those and did all the assignments himself and understood all the concepts, that still puts him at least in the 95th percentile among ML job seekers.
I don't have any ML experience but I don't see what is wrong with it. To me it seems like the equivalent of someone self teaching software development. As long as they are interested and doing a good job there background shouldn't matter much.
Say your company hired a SW engineer who had merely taken an introductory programming course on Coursera, and other than that had no experience. And you immediately make him a senior person, and let him define the role he will play in your company.
Oh, and he didn't have to write any code during the interview.
You don't see anything wrong with that?
I think it's fine to hire a person who just took Coursera ML courses and passes the interview, but you would normally position the person to be a junior with senior folks overseeing his work.
Got a senior ML position in a well known Fortune 500 company. Senior enough that he sets his goals - no one gives him work to do. He just goes around asking for data and does analyses. When he left our team he told me "Now that I have this opportunity, I can actually really learn ML instead of faking it."
If you think that's bad, you should hear the stories he tells at that company. Since senior leadership knows nothing about ML practices, practices are sloppy to get impressive numbers. Things like reporting quality based on performance on training data. And when going from a 3% to a 6% prediction success rate, they boast about "doubling the performance".
He eventually left for another company because it was harder to compete against bigger charlatans than he was.