Amazon interview notes, page 71 of 159

FAANG / Big Tech

Most of these Amazon write-ups come from Software Engineers (1,092), then Data Scientists (182) and Product Managers (165). The most reported rounds are the technical phone screen (618), the behavioral / leadership round (309), the online assessment (228) and the system design round (159).

1581 interview notes · updated Oct 2026

Amazon·Software Engineer·Technical Phone Screen

May 2026

Amazon coding screen focused on log parsing, which sounds boring until you realize the whole thing is basically a trap door of edge cases. The algorithm was never the hard part.

Given a stream of log lines in a structured format, parse each line to extract key-value fields, normalize them, skip any malformed entries, and return a grouped dictionary mapping each key to a sorted list of (timestamp, value) pairs.

“I jumped straight into coding and only realized halfway through that I hadn't confirmed what 'malformed' even meant to them.”

View Post

Amazon·Software Engineer·Technical Phone Screen

May 2026

Amazon SWE coding round with a graph/pathfinding problem disguised as a database table. Pretty clean problem once you see what it's actually asking, but the setup threw me off at first.

Given a table of user navigation records (each row has a user ID, a source URL, and a destination URL), write a function to determine whether a path exists from one URL to another using only the connections recorded in the table.

“Took me a beat to realize this is just a graph reachability problem.”

View Post

Amazon·Data Scientist·Technical Phone Screen

May 2026

Amazon data scientist interview focused heavily on causal inference methodology. One big question that covered a lot of ground, felt like a mini take-home crammed into a verbal answer.

Walk through how you would use Propensity Score Matching and Difference-in-Differences to estimate the causal effect of a marketing campaign. When would you prefer a synthetic control approach over DiD? Also explain the Double Machine Learning framework and why it handles high-dimensional covariates better.

“This was a lot to unpack in one question.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

May 2026

Amazon SWE coding round with a frontend-flavored problem. Not what I expected from them but it was straightforward enough once I stopped overthinking the validation logic.

Build a form with Name, Phone, and Email fields that validates each input and appends a new row to an existing HTML table on success, or shows an error message if validation fails. Vanilla JS only, no libraries.

“Spent way too long debating regex patterns for the phone field.”

View Post

Amazon·Software Engineer·Technical Phone Screen

May 2026

Amazon SWE interview with a coding question around array manipulation. Pretty short session from what I remember, just one problem but it had some edge cases that tripped me up.

Given an array and a series of queries each specifying a range of indices, count the number of items within each compartment defined by those index boundaries.

“I fumbled the first few minutes because I wasn't sure if the compartments were inclusive or exclusive on the boundaries.”

View Post

Amazon·Product Manager·Onsite - Product Sense / Strategy

May 2026

Amazon PM interview with a single estimation question about Zomato delivery drivers in Bangalore. Pretty classic market sizing, but the India-specific context adds some wrinkles if you're not familiar with the local food delivery landscape.

Estimate the total number of Zomato food delivery drivers operating in Bangalore, India.

“I started with population and tried to work down through order frequency, average delivery time, and driver utilization.”

View Post

Amazon·Data Scientist·Technical Phone Screen

May 2026

SQL-heavy technical screen for a Data Scientist role at Amazon. The whole session was basically one big window function problem with a follow-up on indexing strategy, which I did not expect to go as deep as it did.

Given an orders table with 100M+ rows, write a SQL query using window functions to find, for each calendar date (UTC), all merchant(s) whose completed order was the very first of that date across all merchants. Return the date, order_id, merchant_id, and order_ts. If multiple orders share the exact earliest timestamp, return all of them. Do not use correlated subqueries.
Now write a second query: for each merchant and date combination, return that merchant's first completed order of that date.
Justify your partitioning and sorting choices, and describe what indexes you would add to support these queries at scale.

“This is the kind of question where you know the shape of the answer immediately but the details trip you up.”

View Post

Amazon·Data Scientist·Technical Phone Screen

May 2026

Live coding screen for a Data Scientist role at Amazon. One algorithmic problem, no frills, just you and a sorted array and the expectation that you know binary search without leaning on any standard library shortcuts.

Given a sorted array of distinct integers and a target value, return the index of the target if it exists, or the index where it should be inserted to keep the array sorted. You must implement binary search from scratch without using any built-in search utilities.

“I knew the problem immediately but fumbled the pointer logic on my first pass.”

View Post

Amazon·Software Engineer·Online Assessment (OA)

May 2026

Amazon SWE online assessment with a string partitioning problem that looks deceptively approachable until you realize the edge cases pile up fast. Nothing behavioral, just pure algorithmic grind.

Given a string of uppercase letters, for each prefix of length i, determine whether the entire string can be split into consecutive blocks of length i where every block has the exact same 26-letter frequency vector as that prefix. Output 0 if valid, 1 otherwise, for each prefix length.

“The frequency signature idea clicked pretty quickly but I kept second-guessing myself on the edge cases.”

View Post

Amazon·Software Engineer·Technical Phone Screen

May 2026

Amazon SWE coding round with a function timing problem that looks manageable until you realize the exclusive time calculation requires careful stack tracking. Not a brutal interview but the edge cases will get you if you're not careful.

Given a list of function execution logs in the format [function_name, start/end, timestamp], calculate both the inclusive time and exclusive time for a specified target function. Inclusive time is end minus start; exclusive time subtracts the time spent in direct child calls.

“The inclusive part is easy, you just grab the start and end timestamps for the target.”

View Post