you are supposed to answer 10 questions with one minute time given for each. I had 2 -3 questions on concepts of machine learning and rest of them were from concepts of computer science. Things like tuple, duck typing, generator , xrange etc. After that you will have to finish a small machine learning project in your choice of language. Interview process is pretty thorough.
2 technical questions: one machine learning classification problem and one coding question on how to check for a valid BST.
What is the difference between XGBoost and GBDT
Several coding questions across all interviews.
Questions about projects listed on my resume.
What would the classes and methods for a "Hit Point Management" system look like, that has the following functions - increment(user, value), decrement(user, value), getHitPoints(user) and killAll()? The first two methods add/subtract the user's hit points by the amount in value. getHitPoints() returns a user's hit points and killAll() sets all users' hit points to 0. What are the run times of the methods? How can you make all the methods run in O(1)? Note: If we set the hit points data structure to null, or re-initialized it to a new structure, the garbage collector still has to do the work of going through all the users.
Q. How do fasttext embeddings work?
They gave me a machine learning algorithim and asked me how would I derive loss function and why.
What is activation function, and what activation function is used a output layer?
How to use a bayesian framework for a ranking system. SQL queries. Read and explain some code from their codebase.
Viewing 6951 - 6960 interview questions