I had studied a lot of ML but all the technical questions were focused on traditional algorithms, so this was still a little retro. The math requirement was min

I had studied a lot of ML but all the technical questions were focused on traditional algorithms, so this was still a little retro. The math requirement was minimal but I had more than enough from the ML coursework. Bamberg's Math 23a / 23c is phenomenal set of classes at Harvard. You will work like a dog but learn almost all the math you need for ML.