I worked at a non-tech company in the southeast before my current job at a FANG company, and the lead data scientist at this company wrote up the most ridiculous interview questions I've ever seen. WAYY harder than FANG questions, and this company honestly needed business analysts far more than it needed ML researchers. The salary was like $100k for a junior data scientist position. Some of the questions were things like describing forward mode vs reverse mode differentiation, describing Hamiltonian Monte Carlo, and explaining Judea Pearl's concept of causality. Considering there were like 30 data scientists in the whole city, I'm not sure what the point was other than to make potential candidates feel stupid.
I would have told the interviewer to go become a professor and teach his highfalutin causality mumbo jumbo at some university and stop wasting everyone's time.
