The hiring manager was at the interview and first asked me to describe my experience, past projects and things I’m proud of.
Here’s the other two questions I got:
- The engineer puts forward an idea for a new algorithm, saying that the number of upvotes can be increased, and asks how you can check whether it will indeed be improved. (hint - it’s a simple A/Btest)
Then he asked what test tI’d use, I said first use t-test, he said what assumptions, when it comes to normal, he said that it must not be normally distributed. Then I said that otherwise nonparametric or bootstrapped t-test can be used. He continued to ask how to use bootstrapped t-test, I said bootstrapped out two mean confidence intervals and then see if they overlap.
- Now we know that this algorithm has indeed increased upvotes, but found that the number of answers has decreased. What do you think is the reason? . . I thought about this for a while, and said that one of the users tested in this algorithm may have more authority points/ reputation answering questions, so that their answers will have more upvotes. As a follow on assumption I said that people may not leave additional comments once a highly rated user answers since they may trust their answer as right.
Then I got a chance to ask a few questions.
Originally it was supposed to be 45 minutes, but only lasted 30 minutes.