After interviewing alot of candidates, we notices that lot of them where good at solving leetcode like problems but did not understand real world problem. And t

After interviewing alot of candidates, we notices that lot of them where good at solving leetcode like problems but did not understand real world problem. And thats ok I think. You are not supposed to know these things from get go. Lot of the course work is around mathematics, data, machine learning, compilers and that is something that got them excited too. A take home test really works well in this setup as they can research on these concepts and try to solve it if they are really motivated. However this does not scale. We test them on how fast they can learning things, how motivated a candidate is, how driven and ambitious the candidate is. If the grad is really good, these concepts can be picked up pretty fast from the peers.