I see a lot of work done by data scientists and a lot of work done by what I would call “data science flavoured software engineers”. I’ll take the SWE kind any day of the week. Most (not all, of course!) data scientists have an old school “it works on my machine” mentality that just doesn’t cut it when it comes to modern multi-disciplinary teaming. DVCS is the exception rather than the rule. They rarely want to use PMs or UI/UX, and the quality of the software is not (typically) up to production grade. They’re often blinding smart, there’s no doubt about that. But smart and wise are not the same thing.