Amazon Interview Questions — Page 153

Amazon·Software Engineer·Onsite - System Design / Architecture

Jul 2026

Amazon system design round for a software engineer role, one question the whole time about modeling a pizza ordering system. Pretty open-ended which I wasn't expecting, felt like I was rambling for a while before I found my footing.

Design the core object-oriented model for a pizza ordering system. Cover the main classes, their responsibilities, relationships, and how you'd make the design extensible for things like promotions or new menu items. The system needs to support browsing a menu, customizing pizzas, placing orders, calculating prices, choosing pickup or delivery, multiple payment methods, and order status tracking.

“I started with Pizza and Order and thought I was done in five minutes.”

View Post

Amazon·Research Scientist·Technical Phone Screen

Jul 2026

Deep-dive technical screen for an Applied Scientist role at Amazon, focused almost entirely on RLHF training methods. The whole thing was basically a live interrogation on PPO, GRPO, and GSPO with follow-ups that kept going deeper into production realities like rollout cost and reward hacking.

Compare PPO, GRPO, and GSPO. When would you choose one over the others?
How is the advantage estimated in GRPO, and why doesn't it need a value network?
What are the failure modes of reward hacking when fine-tuning a reasoning model, and how do you mitigate them?
Walk me through the rollout cost in RLHF and how you'd handle distributed training topology for GRPO at scale.
Explain the PPO clipped surrogate loss and the role of the KL penalty against a reference policy.

“This was the anchor question and it went longer than I expected.”

View Post

Amazon·Machine Learning Engineer·Technical Phone Screen

Jul 2026

Amazon ML Engineer interview that went deep into recommender systems. The whole session was basically one long question about collaborative filtering, and they kept pushing on the details until I ran out of confident answers.

Walk through how collaborative filtering works for recommendations, and compare user-based versus item-based approaches against matrix factorization. Cover both explicit and implicit feedback scenarios.
What are the tradeoffs between pointwise and pairwise ranking losses in a recommendation context? When would you pick one over the other?
How do you handle regularization in these models, and what goes wrong without it in large-scale recommendation systems?
Describe your approach to the cold-start problem for both new users and new items.
Popularity bias is a known issue in recommendation systems. How do you detect it and what techniques do you use to mitigate it?
How do you evaluate a recommendation model offline versus online? Walk through metrics like NDCG and MAP and explain what each captures.
How do you scale a collaborative filtering system to handle catalogs with millions of items and hundreds of millions of users?

“I started with the neighborhood methods and felt pretty solid there.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

Jul 2026

Amazon SWE online assessment, pretty standard stuff. One coding problem about picking the top-k scores from an array to maximize a sum.

Given an array of subject scores and an integer k, return the maximum possible sum using at most k scores from the array.

“Pretty much just sort descending and sum the first k elements.”

View Post

Amazon·Data Scientist·Technical Phone Screen

Jul 2026

Amazon Data Scientist interview with a coding question built around graph traversal. Pretty algorithmic for a DS role, but not shocking given Amazon's bar.

Given an adjacency list representing a directed acyclic graph, write a Python function that returns the length of the longest path in terms of number of nodes.

“I knew DFS was the right move but fumbled a bit on where to add memoization.”

View Post

Amazon·Software Engineer·Onsite - System Design / Architecture

Jul 2026

Amazon system design round for a software engineer role. The question was a single massive prompt covering basically every layer of input handling you can imagine, from raw key events up through IME composition and crash recovery. Dense but interesting if you're into that kind of thing.

Design an input processing system that captures keyboard and mouse events and can reconstruct what the user typed. The system should support multiple input devices, maintain a total ordering of events, persist an event log, and support replay to derive the final text state.
How would you handle event ordering and timestamps across multiple input devices to ensure a consistent total order?
How would you support undo and redo with sensible transaction grouping, including clipboard operations like cut, copy, and paste?
How would you persist the event log and support crash recovery, including deterministic replay to rebuild the final text state?
What APIs would you expose for this system, for example recording events, querying current text, managing selection, and controlling sessions?
How would you handle robustness concerns like key bounce debouncing, mouse-move event coalescing, and lost or duplicated events?

“This question is enormous.”

View Post

Amazon·Software Engineer·Onsite - Behavioral / Leadership

Jul 2026

Amazon behavioral round for a software engineer role, focused entirely on leadership principles. Two questions, both with aggressive follow-ups that kept drilling into specifics. Not a relaxed conversation.

Walk me through a time you debugged a serious production issue or performance regression. How did you find the root cause, test your theories, and make sure it didn't happen again?
Tell me about a time you took full ownership of a problem where the requirements were unclear or nobody else was stepping up. How did you see it through to the end?

“The follow-ups are where this gets uncomfortable.”

View Post

Amazon·Software Engineer·Technical Phone Screen

Jul 2026

SQL-heavy technical screen for a BI Engineer role at Amazon. Three questions, all centered on window functions and filtering logic. Nothing too wild but the second question had a wrinkle I didn't see coming.

Given a table of seller status events, write a query to find all seller IDs that received a 'block' or 'suspend' status at any point after a 'reinstate' status.
For each seller, return the first status that occurs after their most recent 'reinstate' event.
From a purchase history table, return all customer IDs that bought product 'A' at least once in 2010 and product 'B' at least once in 2018.

“Went with a self-join on the same table comparing dates.”

View Post

Amazon·Software Engineer·Technical Phone Screen

Jul 2026

Amazon SWE coding round, one meaty array problem that took up most of the time. The question had a lot of moving parts and I'm still not sure I nailed the explanation.

Given an integer array, you can optionally pick a contiguous subarray and add any integer (positive, negative, or zero) to every element in it. After this single operation, what is the maximum frequency any value can appear in the resulting array? Also return the subarray range and the delta you used.

“This one wrecked me a bit.”

View Post

Amazon·Data Scientist·Technical Phone Screen

Jul 2026

SQL-heavy technical screen for a DS role at Amazon, focused on logistics/supply chain analytics. One question, but it had enough layers to keep me busy for a while.

For a shipment defect monitoring system, what metrics would you define to track quality on an ongoing basis? Then write SQL that calculates, over the last 30 days, each carrier's total shipments, total defective shipments, and defect rate (defective / total). Return only the top 3 carriers by defect rate.

“The metrics part I actually liked.”

View Post