Background

Status: 4 YOE, currently at a start up
Position: SDE2 at Amazon
Location: Seattle (Remote), WA
Date: May 2022

Phone Screen

Technical
Time: ~20min

1. Given a binary tree, copy it and return the copy.
2. Now given a node, deep copy its children.

Behavioral
Time: ~35min

Onsite

Scheduled onsite for ~1 month after the phone screen

Round 1

Behavioral
Time: ~30min

Technical: OOP Design
Time: ~20min
Question: Design the Linux find command

Round 2

Technical: System Design
Time: ~20min
Question: Design Dropbox

Behavioral
Time: ~30min

Round 3

Behavioral
Time: ~30min

Technical: Data Structures & Algorithms (DSA)
Time: ~25min (Had technical issues, which brought time down to about 10 min)
Question: Design LRU Cache

Round 4 (Bar Raiser)

Behavioral
Time: ~30min

Technical: Problem Solving
Time: ~20min
Question: Something like the following:

You work on a team that processes events across the platform. Implement an API that when called would return the top K most frequent events.

After implementing, a followup was added:

Now suppose your system can only handle 1/10th of the requests coming in, how do you approach this?

Outcome

Rejected for SDE2, given proposal for other available SDE1 roles

Thoughts

I had expected the interview to be much more DSA focused, but that was not the case. If you are prepping it would be good to do more than just pump out leetcode questions as I feel for the most part that did not prepare me as well for this interview. For my experience, the only interview that was labeled DSA (LRU cache) could still be considered by some to be more of a design question (and I have been asked the same question in other roles as a System Design one instead).

Similarly, the final round question was less focused on finding the most optimal solution and was driven more towards how you approached the problem. For example, I provided my thoughts on a few cases of data structures that could be used (such as a heap), and the interview mentioned to not focus so much on the data structures. In the end I provided a solution with simply counting elements and sorting them which he seemed fine with.

Another thing to note was how little time was given for the system design portion. By the time we had gotten through some requirements and discussed some intricacies of one of them, it was time to move on to the behavioral questions. Keep this in mind and if this is your experience as well, pick a part of the design you understand the strongest first so that you can get through it well.

If you plan to interview at Amazon, you should also put a large part of prep into the behavioral questions, as they take these into account a lot more than other companies, and can either lead to rejecting a technically strong candidate or giving a technically weaker candidate a better chance.

Overall, I had a postive experience interviewing and every person I met with seemed to really want to work through the problem together.