Amazon Interview Questions — Page 103

Amazon·Software Engineer·Onsite - System Design / Architecture

Jun 2026

Amazon SWE OOD round. One question, not a lot of warmth from the interviewer's side, which made the whole thing feel a bit stiff.

Design a file system using object-oriented design principles.

“Pretty standard OOD topic but the interviewer was cold the whole time, hard to tell if I was on the right track or completely off.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

Jun 2026

Took the Amazon new grad SDE online assessment and hit TLE on question 1 with 11 out of 15 test cases passing. Wondering if anyone else got through with a similar score.

OA coding problem where 11 out of 15 test cases passed before hitting a time limit error. What was the question and did anyone advance with that score?

“Got 11/15 on q1 due to TLE and now just sitting here refreshing my inbox.”

View Post

Amazon·Software Engineer·Onsite - System Design / Architecture

Jun 2026

Amazon SWE system design round focused entirely on a file-sharing permission control system. Pretty deep dive, they kept pushing on edge cases I hadn't fully thought through.

Design a file-sharing permission control system where users can share files with each other, a homepage lists all files a user can access, and you can check whether a specific user has access to a specific file.
How would you handle consistency after a user's access is revoked? What's the acceptable staleness window and how do you enforce it?
How does your design handle files shared with large groups, say a group with tens of thousands of members?
What changes if your permission model supports explicit deny rules in addition to allow rules?
Walk me through how you'd extend the design to support folder-level permissions that inherit down to files.

“I started with a basic ACL table, which was fine, but the interviewer kept pulling the thread.”

View Post

Amazon·Software Engineer·Onsite - System Design / Architecture

Jun 2026

Amazon SWE system design round, one question about building a distributed logging system. Pretty open-ended, which I wasn't fully prepared for.

Design a distributed logging system.

“Spent the first few minutes just trying to scope it, which I think was the right call but I probably talked too long before drawing anything.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

Jun 2026

Amazon SWE online assessment with two algorithm problems. Nothing behavioral, just pure coding under a timer. The problems were harder than I expected for an OA.

Given an integer array, you can repeatedly pick any contiguous subarray and increment all elements by 1. What is the minimum number of such operations needed to make the array nondecreasing?
Given an array of server memory capacities, pair up servers such that in each pair one server acts as primary and the other as backup (backup capacity must be at least as large as primary). Maximize the total sum of primary server capacities across all pairs.

“This one took me a while to see the right angle.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

Jun 2026

Got an OA link from Amazon for a software engineer role and came here to ask about the AI-assisted coding round since that part is new and I had no idea what to expect. Two coding problems total plus a work simulation, but the AI environment is the part nobody seems to talk about.

How does the AI-assisted coding environment in Amazon's OA actually work, and is it more about prompting skill or raw coding ability?
What kinds of prompts worked well in the AI-assisted coding round, and did you approach it more like pair programming or just using it for debugging?
What are the limitations of the AI tool in the OA environment, and what kinds of tasks did it struggle with?

“Still waiting to take the thing so I can't tell you from experience.”

View Post

Amazon·Machine Learning Engineer·Technical Phone Screen

Jun 2026

Amazon ML Engineer interview that went deep on LLM internals, fine-tuning techniques, and training best practices. Felt more like a technical seminar than a standard screen. The breadth was a lot to cover in one sitting.

Compare encoder-decoder and decoder-only Transformer architectures. Walk through the decoder-only stack in detail, including self-attention, feed-forward blocks, normalization, and residual connections.
How are positional embeddings handled in LLMs? Compare absolute, relative, and rotary approaches and describe the trade-offs.
Describe your hands-on experience fine-tuning large language models. When would you use LoRA versus other adapter-based methods?
During LLM training, how do you prevent overfitting? What regularization strategies, data augmentation techniques, or early stopping criteria do you use?
How do you evaluate a fine-tuned LLM? Walk through both intrinsic and extrinsic metrics, and explain how offline validation differs from online validation.

“This is the kind of question where you think you know it cold and then halfway through your explanation you realize you're fuzzy on why layer norm goes before or after attention in modern implementations.”

View Post

Amazon·Software Engineer·Technical Phone Screen

Jun 2026

Amazon SWE coding round with a tricky array DP problem. The question looked deceptively like a greedy problem at first glance but needed a proper DP formulation to handle the conflict cases correctly.

You have an array of n cities, each with a population value and a binary tag (0 or 1). Tags marked 1 can optionally move one position left or right, but no two tags can occupy the same city. If a conflict arises, you pick whichever placement yields more revenue. Maximize total revenue, where revenue is the sum of populations at cities that end up with a tag of 1.

“My first instinct was greedy, just slide each tag toward the highest neighbor.”

View Post

Amazon·Data Scientist·Technical Phone Screen

Jun 2026

Rapid whiteboard session for a Data Scientist role at Amazon, covering a broad sweep of supervised learning fundamentals in quick succession. The whole thing felt like a pop quiz more than a conversation, which threw me off a bit.

Explain the bias-variance trade-off and how you would diagnose whether a model is suffering from one or the other.
What are the key assumptions behind ordinary least-squares linear regression?
How do you detect overfitting and what steps do you take to address it?
Compare boosting and bagging as ensemble methods, and describe when you would prefer one over the other.
Which evaluation metrics would you select for a binary classifier with heavily imbalanced classes, and why?
How would you handle missing data in your training set?
Why does the normal distribution appear so frequently in statistical modeling?

“I knew this cold but still fumbled the diagnosis part.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

Jun 2026

Amazon SWE interview with a coding problem around memory size calculation for a graph-based model structure, plus a section where you had to actually use AI prompting and debug the output live. Unusual format compared to what I was expecting.

Given a DDN model structure represented as a graph dataset, calculate the total memory size by flattening the graph dimensions and computing size based on the data type's bit size.
Demonstrate how you use AI prompting to solve a coding problem: write the user query, generate code using an AI tool, then debug whatever the AI produces.

“The graph flattening part tripped me up more than I expected.”

View Post