Amazon India SDE Interview Questions for Freshers 2026: Online Assessment, Coding Rounds and Leadership Principles

Amazon India hires fresher software development engineers (SDE-1) through campus drives and off-campus requisitions, and the process is the same shape everywhere: a long proctored online assessment, then a loop of interviews in which every round, including the coding ones, asks about Amazon's Leadership Principles (LPs). Indian freshers usually prepare hard for the coding and treat the LP questions as HR filler; that is the most common reason a candidate with two solved problems still gets rejected. This page explains the 2026 process, then gives 30 questions with sample answers: online-assessment-style problems, interview coding, basic design, and LP questions with examples drawn from Indian college life.

How the Amazon India SDE fresher process works in 2026

As of September 2026, placement-site breakdowns of 2026 drives describe the following. Amazon does not publish a fixed process document; the details below come from candidate reports compiled by PapersAdda, Faceprep and others, and from Amazon's official Leadership Principles page.

Eligibility. B.E., B.Tech, M.Tech, MCA or BCA, with computer science, IT and ECE most common. Placement sites report a common reference of 7.0 CGPA and no active backlogs at joining, set per drive.

Stage 1: online assessment. A single proctored session of roughly 2 to 3.5 hours with three to four parts: a coding section of two data-structures problems, medium to hard, in 70 to 105 minutes; a work-style survey of about 80 Likert-scale statements mapped to the 16 Leadership Principles, 20 to 25 minutes; and in some batches a code-debugging section of 6 to 8 buggy snippets and a logical-reasoning section of series, syllogisms and arrangements, about 20 minutes each. Webcam and screen monitoring are on; tab switching, a second monitor or leaving the frame can flag the attempt. Partial credit is awarded for passing some test cases, so always submit a working partial solution. Inconsistent or extreme answers on the work-style survey can hurt an otherwise strong candidate.

Stage 2: interviews. Three to four rounds, sometimes preceded by a phone screen: coding rounds on data structures and algorithms, a round with basic system design or object-oriented design at fresher depth, and a bar-raiser round with an interviewer from outside the hiring team who focuses on LPs and raises the bar on one coding problem. Every round asks one or two LP questions. On campus, rounds may run on one day; off campus they are spread over one to two weeks on video.

Timeline. Four to eight weeks from application to offer in most reports.

Compensation. Not published. Ask your placement cell for the current breakdown; Amazon offers combine base, signing bonus and restricted stock units, so the first-year and total figures differ.

What the Amazon interviewer is testing

• Correct, working code with the right complexity. Interviewers run your code mentally against edge cases and ask you to improve it.
• Communication while coding. Clarify, state assumptions, narrate, test.
• Leadership Principles with evidence. Specific stories with your action and a measurable result, in STAR form, mapped to the principle being probed.
• Consistency. Bar raisers compare your stories across rounds and with the work-style survey.
• Basic design sense. Classes and responsibilities, or the components of a small system, at fresher depth.

30 Amazon India SDE fresher interview questions with sample answers

Online-assessment-style coding problems

1. Given an array of integers, return the length of the longest subarray whose sum equals k.

"Prefix sums with a hash map from prefix sum to its first index. At each position, if prefix - k is in the map, the subarray between them sums to k; update the maximum length. Store each prefix sum only the first time to keep subarrays longest. O(n) time and space. Handles negatives, which a sliding window would not."

2. Find the minimum number of swaps to sort an array of distinct integers.

"Pair each element with its index, sort by value, then walk the cycles in the permutation; a cycle of length L needs L minus 1 swaps. O(n log n). The intuition is that each swap can fix at most one misplaced element per cycle."

3. Given a string, find the longest palindromic substring.

"Expand around centres: for each index, expand for odd and even length palindromes and track the longest. O(n²) time, O(1) space, which is fine for OA constraints. Manacher's algorithm is O(n) but I would not write it under time pressure unless the constraints demand it."

4. Merge k sorted lists.

"A min-heap of the current head of each list: pop the smallest, push its next. O(N log k) where N is total elements. Divide-and-conquer pairwise merging gives the same complexity. I would mention both and code the heap version."

5. Given a grid of 0s and 1s, count the number of islands.

"Iterate cells; on an unvisited 1, run BFS or DFS to mark the whole island, and increment the count. O(rows × cols). I would use iterative BFS to avoid recursion depth issues on a large grid, and mark visited in place if modifying the grid is allowed."

6. Design a data structure supporting insert, delete and get-random in O(1).

"An array for the values plus a hash map from value to index. Insert appends; delete swaps the element with the last one, pops, and updates the map; get-random picks a random index. All O(1) average. The swap-with-last trick is what interviewers want to hear."

Interview coding rounds

7. Find the lowest common ancestor of two nodes in a binary tree.

"Recursive: if the current node is null or one of the targets, return it; recurse left and right; if both sides return non-null, the current node is the LCA, else return the non-null side. O(n). For a BST, use the ordering to walk down in O(h)."

8. Serialise and deserialise a binary tree.

"Pre-order traversal with a marker for null, joined by commas. Deserialise by consuming tokens with an index. O(n) both ways. The interviewer may ask about handling large trees; I would mention iterative traversal to avoid stack limits."

9. Given intervals, merge all overlapping ones.

"Sort by start, then sweep keeping the current merged interval; if the next starts within it, extend the end, else push and start a new one. O(n log n). Edge case: touching intervals, which I would confirm count as overlapping."

10. Implement a trie with insert, search and prefix search.

"A node with a children map and an end-of-word flag. Insert walks or creates nodes per character; search walks and checks the flag; prefix search walks and returns true if it does not fall off. Each operation is O(length of word). I would mention the memory trade-off of arrays of 26 versus hash maps for children."

11. Detect a cycle in a directed graph.

"DFS with three colours: white unvisited, grey on the current path, black finished. Encountering a grey node means a cycle. Alternatively Kahn's algorithm: if the topological sort processes fewer nodes than the graph has, there is a cycle. O(V + E)."

12. Find the k most frequent words in a document.

"Count with a hash map, then a min-heap of size k keyed on frequency with alphabetical tie-break. O(n log k). Bucket sort by frequency gives O(n) if k tie-breaking is not needed. I would ask how ties should be ordered before choosing."

13. Your solution works. Now the input does not fit in memory. What changes?

"Process in chunks or streams: for counting, hash-partition the input into files by key so each partition fits, count per partition, then merge the top-k results. That is the map-reduce shape. Interviewers at Amazon like to hear you extend a correct solution to scale rather than start over."

Basic design

14. Design a parking lot: classes and responsibilities.

"ParkingLot with levels; Level with spots; Spot with a size enum; Vehicle abstract with Bike, Car, Truck; Ticket with entry time and spot. park(vehicle) finds the first fitting free spot through a strategy interface; unpark(ticket) frees it and computes the fee through a PricingStrategy. Keep allocation and pricing swappable."

15. Design a rate limiter for an API.

"Token bucket per client: a bucket refills at a fixed rate up to a capacity; each request consumes a token; empty bucket means reject with 429. Store state in memory for one server or in Redis for many. I would mention the sliding-window log as an alternative with exact limits at more memory, and ask what the limit and burst requirements are."

16. How would you design the backend for a college canteen ordering app?

"Clients call a REST API; an orders service writes to a relational database; a queue notifies the kitchen display; a cache holds the menu. Payment through a gateway with webhook confirmation. At fresher depth the interviewer wants clear components, the data model for orders and items, and one failure case, such as what happens if the payment webhook arrives twice."

17. Explain how a hash map works internally and what happens on collisions.

"An array of buckets indexed by hash modulo capacity. Collisions go into a linked list or, in Java 8 and later, a balanced tree when a bucket grows long. Load factor triggers resizing and rehashing. A bad hashCode degrades everything to O(n), which is why equals and hashCode must be consistent."

Leadership Principle questions

Use STAR. One story per principle, with your action and a measurable result. The examples below are the shape Amazon interviewers in India respond to.

18. Tell me about a time you went above and beyond for a customer. (Customer Obsession)

"In my internship, users of the attendance app kept mis-marking on slow networks. Nobody had asked me to fix it, but I spent a weekend adding an offline queue that synced later. Support tickets for that issue dropped to zero the following month, and my mentor adopted the pattern in another module."

19. Tell me about a time you took ownership of something outside your role. (Ownership)

"Our final-year project's deployment kept failing the night before the review and the teammate who owned it was unreachable. I read the logs, found a missing environment variable, fixed the script and documented the deployment steps so it would not depend on one person again. We demoed on time."

20. Tell me about a time you had to make a decision with incomplete information. (Bias for Action)

"During a hackathon we had four hours left and two possible approaches. Rather than debate, I proposed building the simpler one for an hour and reassessing. It worked well enough to demo. We placed second; the lesson was that a reversible decision made quickly beats a perfect one made late."

21. Tell me about a time you dug into a problem others had given up on. (Dive Deep)

"A report in my internship showed totals off by a few rupees. Everyone blamed rounding. I compared row by row and found that a currency conversion applied twice for one branch. Fixing it corrected a month of reports. I learned not to accept 'rounding' as an explanation without evidence."

22. Tell me about a time you disagreed with a teammate or senior. (Have Backbone; Disagree and Commit)

"My project lead wanted to store passwords with a fast hash for speed. I explained the risk with a concrete example and proposed bcrypt with a cost that kept login under 200 ms. He agreed after seeing the numbers. When he later overruled me on a UI decision I disagreed with, I committed fully and it turned out fine."

23. Tell me about a time you failed. (Learn and Be Curious; Earn Trust)

"I promised a feature for a demo without checking the API limits of a third-party service. It broke live. I told the team it was my mistake, added a fallback the next day, and started writing a short risk note before committing to any feature that depends on an external service. I have not missed a demo since."

24. Tell me about a time you simplified something. (Invent and Simplify)

"Our team was manually copying data between two sheets every week for a club event. I wrote a script that did it in seconds and shared it; it saved about three hours a week for the organisers and removed the copy errors that had caused double bookings."

25. Tell me about a time you delivered under a tight deadline with limited resources. (Deliver Results)

"Our project review moved up by ten days. I cut two features, split the rest by module, ran a daily fifteen-minute check-in, and we delivered the core on time and scored 9 out of 10. The cut features shipped the following month."

26. Tell me about a time you held yourself or others to a high standard. (Insist on the Highest Standards)

"A teammate's pull request worked but had no tests and duplicated logic. I did not merge it; I paired with him for an hour to add tests and extract the shared function. It took longer that day and saved us when the same logic changed two weeks later."

Wrap-up and logistics

27. Why Amazon?

Say something specific. "Scale: the systems I would work on handle traffic my college project cannot imagine, and I want to learn how that is built. And the ownership model: an SDE-1 owns a service end to end, which is how I learn fastest."

28. Which Leadership Principle do you find hardest?

Honest, with a plan. "Frugality. My instinct is to add a tool or a library for every problem. I have been forcing myself to solve with what exists first, and my last project has half the dependencies of the previous one."

29. What questions do you have for me?

Ask about the team's work and how a new SDE ramps up. "What does the first project for an SDE-1 on your team usually look like?" and "How does the team handle on-call for freshers?"

30. Are you open to any Amazon India location?

"Yes. Bengaluru, Hyderabad, Chennai or Delhi NCR are all fine, and I can join by the date in the offer." Keep the answer consistent with what you said in any earlier round.

Mistakes that get freshers rejected at Amazon India

• Treating LP questions as filler. A candidate with two solved problems and vague stories loses to one with one solved problem and sharp stories.
• Coding in silence. Interviewers grade the thought process as much as the output.
• Not testing your own code. Walk through two edge cases before saying "done".
• Generic stories. "We worked hard as a team" is not an LP answer. Your action, your result.
• Inconsistency between the work-style survey, your stories, and what you told an earlier interviewer.
• Flagged proctoring behaviour in the OA: second monitor, tab switch, leaving the frame.

How to practise the Amazon India SDE fresher round

Two skills decide this loop: coding aloud with edge cases under time pressure, and telling LP stories that survive follow-up questions. Both improve through spoken repetition with feedback, not through reading more problems.

In MockMate Practice, attach your resume and paste the Amazon SDE-1 job description, choose a coding round or a behavioural round and run an adaptive session in the browser. In the coding round you solve a problem in the built-in editor while the interviewer persona asks about complexity and edge cases; in the behavioural round it probes your stories the way a bar raiser does, pushing on what exactly you did and what the result was. The report shows each answer, response timing, coaching evidence and recurring weaknesses, with a recommended next practice. Eligible accounts get three free Practice starts of up to ten minutes each as of September 2026. Use MockMate Live assistance only in interviews or meetings where the organisation or interviewer permits it; Amazon's assessments and interviews are proctored and do not permit it, so stay with Practice for this process.

A two-week plan: week one, one medium problem daily from arrays, strings, trees and graphs, narrated aloud and tested against edge cases; write eight STAR stories mapped to the principles above. Week two, one Practice coding round and one Practice behavioural round every other day, read each report, and rewrite the weakest story or re-solve the weakest problem type. For behavioural question structure see the fresher behavioural page, and for coding-interview preparation the coding interviews use case.