There was an initial phone screen, which lasted for about half an hour, with an HR who asked me basic questions about my background. She tried to discover if I had FE/BE preferences, even though the role was full stack. She was a little vague on compensation, but I chose to proceed.

This was followed by a coding challenge that I did on one of the platforms. It lasted for about 2 hours, and it was considered a good signal if you could get it done sooner (a little weird). They just wanted working code; they didn't care about polish or quality at this stage. I guess they work in a very iterative fashion.

This was followed by an onsite which consisted of 5 parts. I was required to bring my laptop. During the interviews, I would use it to write code, which I thought was pretty good. I was equally comfortable writing solutions on the whiteboard at that stage.

• The first interview consisted of a couple of algorithmic questions (see below). They were not very hard.
• The second was a design review with two people. It was well conducted and interesting.
• The third part was with an engineering manager or director, for whom I didn't feel she was listening to me. She had a laptop, and I'm pretty sure she wasn't taking notes but sending emails/Slack. This felt really bad, and I felt I didn't establish any sort of contact.
• The last part was about database schema design and querying (see below for details).

All the people seemed smart; they were asking good questions, and I felt the bar was high, which is a good thing.

I was rejected with a clumsy and vague email sent from a noreply email address (very poor). I would've thanked them, but... noreply email .

Rating the process as positive because most people seemed smart and polite, barring select few.

The coding challenge involved a function that pulls data from a REST API and then uploads it to another API. There was some data processing in between that required matching, intersecting, and grouping by certain criteria.

Write a function that takes a string of characters and an integer n and returns the most frequent n-gram. For example, given ‘abcdabxe’ and 2, it should return ‘ab’. If there are multiple n-grams with the same highest frequency, return the first one encountered.

Modify this to return the most frequent one.

Given an array of integers and a number n, return true or false if there is a pair that sums to that number.

Design Question: Design a weather widget that pulls data from a meteo service API, which makes data available every hour. The widget should avoid pulling data constantly if there are no changes. Consider what happens when this scales to many users.

Database Questions:

• Social network with Person, Friendship, and Post tables: How would you represent this data? How would you write a query for the wall? How would you implement infinite scrolling?
• Questions about how indexes work, including which index you would create for a specific query.