Amazon Interview Questions — Page 34

Amazon·Machine Learning Engineer·Technical Phone Screen

Apr 2026

Amazon ML Engineer technical screen, one deep question on XGBoost internals that went longer than I expected. The interviewer really wanted to see how far I could go on distributed training specifics, not just the surface stuff.

Walk through how XGBoost parallelizes its training process, covering histogram-based split finding, sparse feature handling, cache efficiency, thread-level work distribution, and how distributed multi-machine training works including synchronization and reproducibility tradeoffs.

“This one went way deeper than I budgeted for.”

View Post

Amazon·Data Scientist·Onsite - Behavioral / Leadership

Apr 2026

Amazon Data Scientist behavioral round, heavy on leadership principles and conflict scenarios. One big question that basically asked you to justify every decision you've ever made under pressure, with numbers.

Tell me about a time you strongly disagreed with a senior stakeholder about whether to launch a data product or model under a tight deadline. What was your role, what were the stakes, and how did it play out?

“This one sprawled in every direction.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

Apr 2026

Amazon SWE online assessment, full-stack debugging task on a movie app. You get a broken recommendations feature and have to find and fix it without any hand-holding from the tooling.

You're given a full-stack movie app where the personalized recommendations section on the home page stays empty even after a user marks a movie as watched and submits a rating. Find and fix the bug.

“The annoying part is the built-in AI assistant only works on a single file at a time, so you can't just scan the whole repo and hope something jumps out.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

Apr 2026

Amazon OA for a SWE role, just one algorithmic problem about tracking when a password becomes irrecoverable as a virus corrupts characters one by one. Pretty niche problem, felt like it was testing interval or segment tree thinking.

A virus attacks a password one character at a time in a given order, replacing each character with a malicious symbol. A password is considered irrecoverable once the number of substrings containing at least one malicious character reaches or exceeds some threshold m. Find the minimum time (in seconds) at which this happens.

“Spent way too long trying to brute-force count substrings after each attack step.”

View Post

Amazon·Software Engineer·Technical Phone Screen

Apr 2026

Amazon SWE coding round, pretty much just one meaty question that kept expanding. Started with 2Sum, then they pushed it to 3Sum, and they wanted the full treatment: walkthrough, code, dry run, edge cases, and complexity analysis.

Given an integer array and a target value, return all unique pairs that sum to the target. Pairs should be sorted and contain no duplicates. Walk through your approach, write a complete Python solution, dry run a test case, list unit test cases, identify an edge case your solution can't handle, and analyze time and space complexity.
Follow-up: extend your solution to find all unique triplets instead of pairs. How does the approach change, and what are the new time and space complexities?

“The 2Sum part felt fine.”

View Post

Amazon·Software Engineer·Technical Phone Screen

Apr 2026

Amazon SWE interview with a heavy focus on interval-based problems. Felt like they had a whole theme going with this category, covering everything from merging to scheduling to maximizing non-overlapping sums.

Given arrays of start times, durations, and costs for a set of intervals (some of which may overlap), find the maximum total cost achievable by selecting non-overlapping intervals.
Find the total overlapping time across a collection of intervals.
Given a list of intervals, merge all overlapping ones and return the resulting set of non-overlapping intervals.
Given a list of tasks with cooldown constraints, find the minimum time needed to complete all tasks (Task Scheduler variant).
Find the optimal difference between intervals given some set of constraints.

“This is the one that actually made me think.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

Apr 2026

Amazon SWE online assessment with a log analysis problem involving interval queries. Nothing too wild but the edge cases in the problem statement are easy to miss if you're moving fast.

Given a list of skill IDs and timestamps representing request logs, and a time window, for each query time determine how many skills received zero requests in the interval [queryTime - timeWindow, queryTime]. Return an array of counts.

“The problem reads straightforward until you realize you need to handle skills that never appear in the logs at all.”

View Post

Amazon·Software Engineer·Onsite - Behavioral / Leadership

Apr 2026

Amazon SWE behavioral loop, all leadership principle style questions. Six prompts total, pretty standard for the company but the generative AI one threw me a bit since I wasn't expecting to talk product workflow in a software eng interview.

Tell me about a time you were working against a tight deadline. Did you miss it, and what happened?
Describe your experience using generative AI tools. How did you validate the output, and what would you change about the workflow or the product?
Tell me about something you learned on your own that wasn't part of your job responsibilities.
Walk me through a time you had to dig deep into a complex problem. Why did you approach it the way you did, and what would you do differently now?
Tell me about a time your work had a significant impact on other teams. What other options did you consider, and how did you handle disagreement about the direction?
Describe a time you worked outside your comfort zone. What did you take away from it, and what would you do differently?

“The follow-up about whether I actually missed it is the part people forget to prep.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

Apr 2026

Amazon SWE online assessment with a DP/state compression problem involving bundle pricing. The problem was a variant of the shopping offers type where you have to figure out the minimum cost given unit prices, special bundle deals, and required quantities.

Given a list of item prices, a set of bundle offers (each specifying quantities of items and a bundle price), and a required quantity for each item, find the minimum total cost to fulfill the requirements. You can use each bundle any number of times and also buy items individually.

“This is basically a state compression DP where each state is the current quantity vector of items you still need.”

View Post

Amazon·Data Scientist·Technical Phone Screen

Apr 2026

Amazon data scientist interview with a stats-heavy question about p-values in the context of A/B testing. One question, but it had a lot of surface area to cover and I don't think I hit all of it.

You're running an A/B test for a new product feature and stakeholders are using p-values to make the ship decision. Walk through at least three real limitations of p-values in this setting and what you'd recommend instead or alongside them.

“I started with sample size dependence because it felt safest, basically that with enough users you can get a significant p-value on a completely trivial effect.”

View Post