The interviewer was very respectful and assessed my knowledge about distributed systems, cloud, and deployment tools.

I was asked about a challenging project I worked on in the past. I think this kind of question is highly expectable.

I highly appreciate being challenged regarding my technical background as well as my thought process. My expectation was to go through an actual, real-world-like scenario. I'm happy being asked questions about stuff that I will actually work on, but I was challenged with some obscure data structure problems, which frankly say nothing about how I work.

For example, I couldn't showcase my knowledge about the development and deployment of machine learning models at scale. This is super important, and there are not many people in the industry with real experience.

Perhaps this is the way Twilio conducts interviews. With all due respect, I couldn't disagree more.

Imagine you have a bag of marbles. This bag would contain, say, 20 red marbles, 10 green marbles, and 5 blue marbles.

We are going to create software that represents this bag. The main operation we need to be able to do is to repeatedly draw a marble at random from this bag without replacement.

Draw up a data structure for how you represent this bag and describe how the draw operation works, such that the bag, both before and after the operation, properly statistically represents its contents.