Updated by Amazon candidates
Amazon L3 Interview Experience
They brought up on-call in multiple conversations, so that was the culture signal I actually believed. Then the hiring manager had me design a package tracker that could survive duplicate, out-of-order updates during Prime Day.
Interview process
I went through four rounds: a 90-minute OA, a live coding plus behavioral round, a tree plus Leadership Principles round, and then a hiring manager final with light system design. The OA was two coding questions, then the live rounds stacked coding with behavioral in a pretty straightforward way. Most of it landed around medium difficulty for me. The round that felt hardest was the tree plus Leadership Principles one because that was where the optimization follow-ups and deeper probing really picked up. The final round was more of a high-level package-tracking design discussion than a super deep system design.
- Online assessment
- Technical interview
- Final round
Interview rounds
8 questions across 4 roundsOnline assessment
2 questionsOverview
I started with a 90-minute OA that felt like pretty standard LeetCode Medium. It was two coding questions, and the main thing was moving fast but still catching edge cases and performance issues.
Question types asked
Specific questions asked
Solve an arrays and hash maps problem.
I finished this one fairly quickly. It was mostly arrays and hash map logic, and I made sure to cover edge cases and keep it efficient for large inputs.
Solve a sliding window optimization problem.
This one took me longer because I had to debug part of my logic. I got to the optimized sliding window approach, but the harder part was making sure it handled edge cases and large inputs cleanly under time pressure.
Technical interview (1)
2 questionsOverview
The next round was a 60-minute technical that mixed live coding with a behavioral discussion. It felt medium difficulty, and the interviewer cared about both the algorithm and how I explained tradeoffs.
Question types asked
Specific questions asked
Find the shortest path through a grid with movement restrictions.
What are the time and space complexities of your approach?
How would your solution scale?
I started with a brute-force idea, then talked through why BFS was the better fit for shortest path in a grid. I explained the transition from the naive approach to BFS, then walked through the complexity and how it would behave as the grid got larger.
Tell me about a previous engineering project and what you personally contributed.
What specifically was your individual contribution?
I walked through one of my past engineering projects and kept the focus on what I owned directly. They wanted to separate my work from the broader team effort, so I made sure to be clear about the decisions I made and what I implemented myself.
Technical interview (2)
2 questionsOverview
This round was split between a tree coding problem and Leadership Principles, and it was probably the toughest round for me. The coding part had the usual optimization follow-ups, and the behavioral half had a lot of drilling into Ownership and Dive Deep.
Question types asked
Specific questions asked
What are the time and space complexities?
Can you optimize your initial approach?
How would your solution handle negative node values?
I first described the brute-force recursive version, then moved to the optimized DFS approach using a prefix sum hash map. I explained the recursion before coding, walked through edge cases, and talked about why negative values matter because you cannot rely on pruning the way you might in a simpler path sum variant.
What exactly did you contribute?
Why did you make those technical decisions?
What impact did your work have?
This was very much an Ownership and Dive Deep story. I framed it around spotting a technical issue other people had missed, then they kept pushing on what I personally did, how I chose the fix, and what changed because of my work, so I had to stay very specific instead of giving a team-level answer.
Final round
2 questionsOverview
The last round was a 60-minute hiring manager interview that was more high-level than a full deep system design. It centered on an Amazon-flavored package-tracking problem plus some behavioral questions around technical decision-making and Leadership Principles.
Question types asked
Specific questions asked
How would you support millions of packages?
How would you handle frequent status updates from warehouses and delivery drivers?
How would you make sure customers see accurate tracking information with minimal delay?
How would you deal with duplicate or out-of-order updates?
What APIs would you expose?
How would you choose the database?
How would the system scale during peak events like Prime Day?
I treated it as a high-level design discussion. I talked about supporting millions of packages, ingesting frequent updates from warehouses and drivers, and keeping tracking info accurate with low delay. I proposed using a message queue for asynchronous processing, then discussed how to think about duplicate and out-of-order events, database choice, and scaling traffic during Prime Day.
Interview tips
I would prep for standard LeetCode Medium patterns more than anything fancy. Arrays and hash maps, sliding window, BFS on grids, and tree DFS with prefix sums all showed up for me, and they definitely cared about whether I could explain complexity and improve a brute-force idea instead of just landing on code. For behavioral, I would have Ownership and Dive Deep stories ready and keep them very concrete, because they pushed hard on what I personally did, why I made those technical decisions, and what changed because of my work. For the hiring manager round, I would practice a light system design where the real discussion is around event handling, like duplicate or out-of-order updates, async processing, and scaling for peak traffic.
Company culture
This felt pretty standard compared to other interview processes I have been through. The thing they kept reinforcing was team culture, and specifically that the team is on call frequently. That came up across multiple conversations so it must be legit. The vibe from the people I talked to was good.
Get full access with a membership, or share your experience to try it free.
More Amazon interview experiences
“It was the usual four-round AWS loop, but the HM actually spent time on how I use GenAI to fast track my workflow, and the Bar Raiser was basically all LPs with tons of follow-ups.”
“I went in expecting the usual behavioral-first format, but in my loop they flipped it and opened with the coding question. The behavioral part still mattered, they just used the follow-ups to see if I actually knew my own work in detail.”
