Meta Data Scientist interview questions

If you had to design a feature for YouTube, what would it be and how would you test it?

1- 3 to 4 SQL question from the employee department with join, and conditional selecting 2- Stat question(OLS, assumptions, t-test, F-test in OLS? Are they both necessary or redunt? what metrics can you use, R squared and probability, what are the problems with R2? How does maximum likelihood relate to ols? Under what condition they are the same? If it possible by adding a feature r2 to decreased? 3- Machine learning: logistic regression vs linear regression, what is decision tree, how does it select feature? How does it decide to which feature to select? random forest what is the 2 element of randomness? What is the kernel trick in SVM? 4- Given a list and a target value, how to find the 2 element that the sum of them is equal to target? Space and time complexity? How to improve it? How the dictionary works?