Amazon Interview Questions — Page 107

Amazon·Software Engineer·Online Assessment (OA)

Jun 2026

Amazon SWE online assessment with an interval-overlap problem dressed up in marketplace language. The core idea isn't too bad once you strip away the flavor text, but the wording tripped me up more than I'd like to admit.

Given n merchants each with an operating zone defined by a start and end point, find the minimum number of merchants that must be removed so that the remaining group has at least one merchant whose zone overlaps with every other merchant's zone.

“The problem wraps an interval scheduling question in a bunch of marketplace flavor text and I spent way too long parsing what 'cohesive' actually meant.”

View Post

Amazon·Software Engineer·Onsite - Coding / Algorithms

Jun 2026

Amazon SWE coding round, one algorithm question the whole time. Pretty standard if you've prepped graph problems, but the specific variant they picked tripped me up a bit.

Given a list of courses and their prerequisites, determine if it's possible to complete all courses (a topological sort / cycle detection variant).

“Knew it was a graph problem pretty fast, but I fumbled the cycle detection part.”

View Post

Amazon·Data Scientist·Technical Phone Screen

Jun 2026

Amazon data scientist interview that went deep into statistical ML fundamentals. Five meaty questions covering everything from OLS derivations to AdaBoost weight updates. Not a vibe check, they wanted actual math.

Derive ordinary least squares from scratch: walk through the model setup, assumptions, normal equations, the closed-form estimator, when the matrix inverse exists, the ridge regression alternative, and how regularization shifts the bias-variance tradeoff.
For logistic regression: write out the negative log-likelihood for binary classification, derive the gradient and Hessian, prove the loss is convex, then do one explicit gradient descent step with learning rate 0.5 using x=(1,2), y=1, and current weights w=(0.1,-0.2).
Name three distinct techniques to combat overfitting, explain when each one helps versus when it could hurt performance, and then design a cross-validation strategy to tune the L2 regularization strength lambda.
Bootstrapping vs boosting: for bootstrapping, describe the percentile interval method for estimating uncertainty on a mean using the sample [2,3,5,7,11], show two example resamples with replacement and their means, and explain why this works without parametric assumptions. For boosting, explain the core idea and then manually run one AdaBoost iteration with three equally-weighted points where the weak learner misclassifies only the second point, computing epsilon, alpha, unnormalized weights, and the renormalized distribution.
Compare bagging, boosting, and random forests across bias, variance, and sensitivity to noisy labels. Give a concrete scenario where you'd prefer each approach.

“This took way longer than I expected.”

View Post

Amazon·Software Engineer·Onsite - Behavioral / Leadership

Jun 2026

Behavioral round at Amazon for a solutions architect role, pretty much what you'd expect from them. One question, classic leadership principle territory.

Tell me about a time you weren't able to follow through on a commitment. What happened as a result?

“This one stung a bit because I had a real story but fumbled the delivery.”

View Post

Amazon·Software Engineer·Onsite - System Design / Architecture

Jun 2026

Amazon SWE design round focused entirely on building a rate limiter. The prompt was intentionally vague and the real test was whether you'd drive the requirements conversation yourself before writing a single line of code.

Design a rate limiter for inbound user requests. Implement an allow(user_id, timestamp) method that returns a boolean. Time window and request limit are not specified upfront.
Walk through the trade-offs between fixed window, sliding window log, sliding window counter, and token bucket implementations.
How would you handle thread safety in a single-node rate limiter implementation?
How would you extend the rate limiter to work across multiple distributed nodes?
What happens to your rate limiter if the shared state store goes down? How do you handle that failure mode?
How would you deal with clock skew across multiple gateway nodes in a sliding window implementation?

“The sparse prompt is the whole point.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

Jun 2026

Amazon SWE online assessment with a greedy array problem about relocating security units to maximize protected population. Pretty straightforward once you see the pattern, but the edge cases in the examples took me a minute to work through.

Given an array of city populations and a binary string indicating which cities have a security unit, you can move any unit one step left (at most once per unit). A city is protected if it has a unit after all moves. Return the maximum total population of protected cities.

“My first instinct was greedy from left to right, which was wrong.”

View Post

Amazon·Data Scientist·Online Assessment (OA)

Jun 2026

Amazon OA for a Data Scientist role. Two parts: task prioritization and a recorded coworker response. Not the most technically demanding assessment I've done, but the soft-skills angle tripped me up more than I expected.

You're given a set of competing tasks with different deadlines and customer impact levels. Rank them by priority and explain your reasoning.
After listening to a short audio clip of a coworker describing a problem, record your verbal response addressing their concerns.

“I went straight to deadline urgency and kind of forgot to weight customer impact properly until halfway through my justification.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

Jun 2026

Amazon SWE online assessment, one algorithmic problem about allocating extra time across tasks to maximize completions. Pretty clean greedy problem once you see it, but I second-guessed myself on the sorting logic for a bit.

Given two arrays representing the required time and already-spent time for each task, plus a fixed budget of extra time you can distribute freely, return the maximum number of tasks you can complete.

“The key is computing remaining time per task and then sorting by that ascending, then greedily picking the cheapest tasks until you run out of budget.”

View Post

Amazon·Product Manager·Onsite - Behavioral / Leadership

Jun 2026

Amazon PM loop, behavioral round focused on how you handle recurring operational friction within the team. One question, but it had a few layers to it and they clearly wanted to see structured thinking about communication and trade-offs.

A teammate who owns critical data is consistently late to a bi-weekly meeting because it overlaps with another call they have. Walk through how you'd choose the right way to address it with them, what options you'd weigh (reschedule, collect data ahead of time, send someone else), and how you'd land on a solution without tanking team morale or slipping on delivery.

“I spent too long on the communication channel part, like whether to call vs message vs walk over, and kind of buried the actual solution.”

View Post

Amazon·Software Engineer·Technical Phone Screen

Jun 2026

Amazon SWE coding round, one question about grid traversal and robot coverage. Pretty standard algorithmic problem but the details matter more than you'd think.

You're given a 2D grid representing a factory floor. Some cells are blocked, others contain robots. Can the robots collectively reach every unblocked cell? Robots move in four directions and can't pass through blocked cells.

“My first instinct was multi-source BFS starting from every robot simultaneously, which is the right move.”

View Post