Comment by kriro

Interesting, thanks. It obviously depends a lot on the job but I would have thought that a good math background (which typically CS grads should have) would be enough for many applications. Especially since you mention the business problem aspect. I think that you can actually get quite far in DS (at least in deep learning) with rather "basic math". Unless you're innovating at the bleeding edge it's usually enough to reproduce existing ideas and applying them to interesting problems is the real key. So my hiring would have focused on CS grads with a natural interest in math, ideally very rigorous/meticulous.