Netflix Machine Learning Mock Interview: Type-ahead Search 

His answers are very random

The positive comments in the comments section are extremely baffling. This is one of the worst ML Design answers I've ever heard in my life.

Jay, this is awesome please don't stop making these videos!!

Totally random flow of thought..Nothing makes sense till 13 minutes where he mentions Trie. K-means clustering to get Coen brother fan, prefix table with all possible strings - these things show that you don't understand ML at all.

There was barely any ML related discussion in the answer No discussion on metrics or evaluation When interviewer pointed out sparsity of user profile, the answer was surprisingly worse than original There wasn't any discussion on once you get candidates matched from Trie, how would they score it using user profile When asked on scalability, the answer was far too focussed on the feature set scalability

Jay - I would like you to be the person telling answers in some cases. Your knowledge is very good and would love u see this in parallel with the guests

Those boardgames though...

Hi, could you elaborate on 12:02 feature expansion? A little lost on Dan's explanation. Who decides on which new feature to include out of the thousands? thanks

Hi, Is this supposed to be an interview for an entry level job or a senior level one? Great job on the videos either way!

8:00 bro, that'll help develop the priors for his Max likelihood estimate

Thanks Jay, where do you find these interview questions from big tech companies?

This is amazing .

Wow the guy answering is very very good in algorithms and broad spectrum