>>55360047
No, due to reliability. You can know pretty much everything about it. But if something goes wrong or if you do something wrong. You'll be blamed because you don't have a degree. Which is a risk.

>>55360047
No. Even with only a master's it would be sketchy as fuck and require lots of connections.

Maybe. Blog posts like https://blog.intercom.io/machine-learning-way-easier-than-it-looks/ indicate companies are open to the self-educated as long as they deliver. I think this will especially be the case in "hybrid" ML/dev jobs.

>>55360047
Yes, DS I work with has a physics undergrad, but that's it. Came in as an intern on that team while the company was a small startup and worked his way up.

The rest of them have PhDs, though.

You can be a data analyst without one no problem, but DS is a major meme atm, so hence the scramble for PhDs.

As this guy points out >>55361557, a lot of the memier shit isn't actually rocket science. Anyone with a decent grasp of python can get cracking on NLTK for ML, can use something like Pandas for fucking around with data, and can learn to use things like Tweepy for pulling Social Media stats.

Great article I read (can't find sauce now) argued convincingly that while businesses go for DS because it's sexy atm and they think they need it, in practice what they're actually looking for is more like a Business Intelligence person, who can do data analysis, but can also figure out how to implement shit the business needs- cool stuff like ML is pretty secondary to that.