Amazon | SDE 2 | Loop | USA - Reject
Summary
I completed a full loop interview for an SDE 2 position at Amazon and despite performing strongly in system design and some coding rounds, I was ultimately rejected. They did mention considering me for a Technical Program Manager role due to my system design performance, and my reapplication cooldown period was reduced to six months.
Full Experience
I completed a full loop interview for an SDE 2 position at Amazon in February 2025. I have just under 3 years of experience at another MAG7 company. Here's a breakdown of my experience, excluding the Leadership Principles (LP) discussions:
Logical and Maintainable - Low-Level Design Interview
The problem was to design a Grocery Cart Checkout system. The core requirements were that a food item has a list of ingredients, the cart needs to check for ingredients that add extra taxes, and the total should account for both basic tax and ingredient-specific taxes.
Result: I didn't fully finish this round because I was initially confused, thinking it might be a LeetCode-style coding interview rather than an OOP-focused design. Despite this, I was about 90% complete with my design. The LP questions asked in this round went very well.
System Design - High-Level Design Interview - Bar Raiser
I was asked to design an Ad Click Aggregator. The system needed to scale up to 10 million clicks per minute. Advertisers should be able to query their click data in real-time, with latency under 2 minutes. Additionally, customers should be redirected to ads with minimal latency, specifically under 200ms.
Result: I felt I did great in this interview; it was my best performance. I struggled with one particular deep dive question but handled another very well. The LP discussion for this round also went amazingly.
Data Structures and Algorithms Interview - Hiring Manager
This round involved a problem that was a modification of the Rotting Oranges LeetCode problem. The twist was that instead of squares, the 'cells' were hexagon shapes, meaning 6 directional spreading. It was framed as a wildfire simulator where brushes spread the fire. To add complexity, there was directional windspeed that could cause the fire to spread at a longer, variable distance in one specific direction.
Result: One LP question went great, though I didn't seem to fully satisfy the interviewer on another. My coding performance for this problem was excellent, and the interviewer seemed very happy with my solution.
Problem Solving Interview
I was presented with two problems, both based on the Jump Game series from LeetCode. The first was Jump Game and the second was Jump Game II. The specific modification for both was that instead of outputting just whether a jump path exists or the total number of jumps, I had to output the actual jump path itself.
Result: The LP questions for this round went great, and the interviewer was very happy. I finished the first Jump Game problem without any issues. After that, the interviewer asked three non-coding follow-up questions on the first Jump Game, which I answered to their satisfaction. However, I wasn't able to finish the second Jump Game problem and struggled with it during the interview.
Overall Result
I ultimately received a rejection, falling short just in the coding portion. Because I did well in system design, they mentioned they would consider me for a Technical Program Manager role if I was interested. They also said that due to my relatively good performance, my cooldown period to reapply would be 6 months instead of a year.
Interview Questions (5)
Design a Grocery Cart Checkout system. A food item has a list of ingredients. The cart should check for ingredients that add extra taxes. The total should account for basic tax + ingredient taxes.
Design an Ad Click Aggregator system that can scale up to 10 million clicks per minute. Advertisers should be able to query their click data in real-time (latency < 2 minutes). Customers should be redirected to ads with minimal latency (< 200ms).
A variation of LeetCode's Rotting Oranges problem. Instead of squares, the 'cells' are hexagon shapes (6 directional spread). It's modeled as a wildfire simulator with brushes spreading the fire. Additionally, there is directional windspeed that can make fire spread at a longer, variable distance in a certain direction.