All of the questions can be answered with standard joins, aggregations, loops, and dictionaries and arrays. You guys are way over complicating it with nTile sql statements and pandas libraries.

For SQL there are steps 1,2, and 3, and a time stamp. You can deduce time spent on each step By subtracting timestamp 1 from 2, then averaging the differences for average time spent on each step. For python, as new times stream in for each user (user iD and step number and timestamp) find out if we’ve already seen this user and if so find the difference between step time stamps, add the difference to a running total, increment the “seen” users count by 1, and divide total by number of users to be able to return the average page loading times on demand

Thank you for sharing you experience! I am quite curious to know about the study material - they give. From your experience do they ask questions around the case studies, that your are asked to prep for each round or you were asked questionns on different subjects/ topic/ products? Will be helpful to learn about your experience.

Lol that makes sense. Thanks for clarifying.:)

Was this the only full stack question they asked? There are generally 3 full stack rounds, can you share about other rounds as well.

it says multi step product feature, I don't think it should be time series of performance.