Amazon Interview Questions — Page 131

Amazon·Software Engineer·Onsite - System Design / Architecture

Jun 2026

Amazon system design round focused entirely on building a configuration service from scratch. It went deep fast, covering everything from basic API design to distributed fan-out and rollback semantics. Not a casual question.

Design a configuration service that lets applications set and get config values, subscribe to specific keys or namespaces, and receive notifications when values change. Define the API surface including set, get, subscribe, and unsubscribe.
What are the notification semantics for your config service? Specifically, how do you handle at-least-once vs at-most-once delivery, message ordering, and whether to debounce rapid successive changes?
How would your config service handle versioning, rollbacks, and partial failures during a config update?
How do you ensure thread safety and handle concurrency in the config service?
How would you handle persistence and backup for the configuration data, and what access control mechanisms would you put in place?
How would you scale this config service to thousands of clients? Walk through your fan-out strategy, caching approach, and your choice between long polling and WebSockets for client communication.
What is the time and space complexity of the core operations in your config service, specifically get, set, and subscribe?

“I started with the API shape which felt natural, but I underestimated how much they'd push on subscription granularity.”

View Post

Amazon·Software Engineer·Onsite - Behavioral / Leadership

Jun 2026

Behavioral round at Amazon for a software engineer role, pretty much one question that took up most of the conversation.

Describe a time you had a conflict with a teammate or colleague. How did you handle it, and what was the outcome?

“Picked a story about a code review disagreement that dragged on longer than it should have.”

View Post

Amazon·Data Scientist·Onsite - Behavioral / Leadership

Jun 2026

Amazon Data Scientist behavioral loop, two questions deep into leadership principles territory. Both prompts were pretty demanding in terms of specificity, way more than I expected for what I thought would be a softer screen.

Tell me about a time you radically simplified a complex workflow by inventing a new tool or process. Walk me through the before and after, the trade-offs you considered, the hardest constraint you faced, risks you had to manage, and what the measurable impact was. Also, how did you get skeptics on board, and what would you do differently?
Describe a situation where a miscommunication with an external customer put a project outcome at risk. How did you figure out where the gap was, get everyone using the same language, confirm you were actually aligned, and manage disagreement when time was short? Include a concrete example of how you'd communicate that alignment in writing.

“This one bit me because I had a decent story but I kept getting pulled into the weeds on technical detail instead of landing the numbers early.”

View Post

Amazon·Software Engineer·Technical Phone Screen

Jun 2026

Went through Round 1 for an Amazon SDE 1 position, which covered one behavioral question and one coding problem. Felt shaky in spots but came out thinking it was probably a passing performance, though I genuinely can't tell.

Tell me about a specific experience from your past work or projects that demonstrates one of Amazon's core leadership principles.
Solve a coding problem requiring an optimal algorithmic approach, then walk through your solution's time and space complexity.

“Nerves hit hard at the start but faded as the conversation went on.”

View Post

Amazon·Software Engineer·Technical Phone Screen

Jun 2026

Amazon SWE interview, one deep-dive debugging question that took up basically the whole session. The kind of question where you think you know what they want and then realize halfway through they want something completely different.

Walk me through a time you had to dig deep into a system to fix a hard problem. How did you instrument it, form hypotheses, look through logs and metrics and code, and confirm the root cause? What trade-offs did you make along the way, and what did you change long-term?

“The part I fumbled was the trade-offs section.”

View Post

Amazon·Software Engineer·Technical Phone Screen

Jun 2026

ML breadth round for an Applied Scientist role at Amazon, no coding involved. They covered a lot of ground on classical ML theory and threw in one leadership principle question at the end. Felt more like a grad school oral exam than a typical industry interview.

Walk me through logistic regression: derive the loss function, explain why cross-entropy is used, and connect it to maximum likelihood estimation. Also discuss L1 vs L2 regularization.
Explain how CNNs work: cover the convolution operation, pooling, receptive fields, parameter sharing, and how skip connections help with vanishing gradients.
Explain the bias-variance tradeoff and how it relates to overfitting and underfitting. What techniques do you use to address each?
When would you use different loss functions? Walk through your reasoning for picking a loss given a specific problem setup.
How do you think about evaluation metrics like precision, recall, F1, and AUC? When does each matter more than the others?
Tell me about a time you found a simpler solution to a problem that others were overcomplicating, or introduced something new that eliminated a recurring issue entirely.

“This one went long.”

View Post

Amazon·Machine Learning Engineer·Hiring Manager Screen

Jun 2026

Hiring manager screen for an MLE role at Amazon. The whole thing revolved around walking through a recent project in depth, with distributed systems and leadership principles woven in throughout. More conversational than I expected but also more probing.

Walk me through a recent ML project end-to-end: what motivated it, what technical decisions you made, how the system was architected, and specifically what you contributed.
What distributed systems challenges did you encounter in this project, and how did you handle them?
How would you address gaps in your distributed systems background if this role requires deeper expertise in that area?

“This was the bulk of the interview.”

View Post

Amazon·Software Engineer·Technical Phone Screen

Jun 2026

Amazon SWE interview with a data structures question that went deeper than I expected. The circular queue problem sounds straightforward until you're actually explaining index management out loud under pressure.

Design a fixed-capacity circular queue using an array. It needs to support enqueue, dequeue, front, rear, isEmpty, and isFull, all in O(1). Walk through how you handle the head and tail indices, how you tell empty from full, and what the time and space complexity looks like.

“I knew the modular arithmetic part well enough but fumbled explaining how to distinguish empty vs full state.”

View Post

Amazon·Data Scientist·Technical Phone Screen

Jun 2026

Amazon BIE-style SQL interview for a Data Scientist role, focused on relational fundamentals. Three questions, all conceptual, no actual query writing required which surprised me a bit.

What is the difference between DELETE and TRUNCATE, and when would you choose one over the other?
How does a VIEW differ from a physical table, and what are the tradeoffs of using one versus the other?
What does a CROSS JOIN produce, and can you give a realistic scenario where you'd actually use one?

“I knew the basics but fumbled the transactional logging part.”

View Post

Amazon·Software Engineer·Technical Phone Screen

Jun 2026

Amazon SWE technical phone screen, one question, pure algorithms. They wanted the full treatment: correctness argument, complexity, edge cases, pseudocode. Pretty standard BFS grid problem but they pushed hard on the details.

Given an m x n grid where -1 marks an obstacle and all other cells are passable, find the shortest path from the top-left to the bottom-right corner using only four-directional moves. Return -1 if no path exists. Walk through your algorithm, argue its correctness, analyze time and space complexity, handle edge cases like a blocked start or goal or a single-cell grid, and write pseudocode.

“BFS is the obvious call here and I got there fast, but then they kept pulling on threads.”

View Post