How to Crack Amazon: The Software Engineer Interview, From Real Experiences
Prepare for Amazon software engineer interviews with reported assessment patterns, interview questions, and preparation advice.
On this page
Last reviewed November 15, 2026.
For interns, new grads and experienced software engineers.
Built by the Codinzhub team from Amazon interview experiences that candidates shared, including the ones posted on Codinzhub. Each experience is one person's interview, and Amazon changes its process by year, level and team, so use this as a map, not a promise.
The 30-second version. An online assessment in three parts: a hard coding question, an AI-assisted debugging task in a real project, and a work-style survey built on the Leadership Principles. Then sometimes a phone screen, then a loop of three to five one-hour rounds on one or two days. Every round mixes Leadership Principle (LP) questions with coding, one round is a low-level design, SDE II and above add system design, and one interviewer is the Bar Raiser. Candidates said LP time can exceed coding time.
Round-by-round breakdown
Stage 1: Application, where candidates applied
How it works. Most candidates applied on the careers site without a referral, often months before the OA arrived (January to April in one case). Others were contacted by HR on LinkedIn, sometimes several times. Campus drives and Amazon HackOn led to SDE intern and 6-month intern roles at IIT BHU, IIT Roorkee, IIT Patna, IIIT Allahabad, IIIT Delhi, DTU, NIT Jamshedpur, NIT Bhopal, MNNIT, ISM Dhanbad, BITS Pilani and others.
Stage 2: Shortlisting, how candidates moved forward
How it works. The OA decides who gets interviews. New grads often had no recruiter call: an email from Amazon University Talent Acquisition asked for availability, and the loop came a week later. Experienced candidates had a recruiter call, then a phone screen. One candidate's OA passed in mid-June after applying in January. Roles can be filled while you wait; HR then offered other teams.
Stage 3: Online assessment
At a glance
- Length: about 2 to 3 hours in parts: 40 to 70 minutes of coding, 60 minutes of AI debugging, then surveys
- Questions per round: 1 or 2 coding questions, 1 debugging project, a work simulation and a work-style survey
- What was asked: sliding windows over huge ranges, greedy with bits, DP, fixing a Django, Node or Java app with an AI assistant
- Most common: the maximum coins in k consecutive bags; fixing failing tests in a web app
How it works. The coding part is usually one hard problem with hidden tests. The AI debugging part gives you a small full-stack app (Django, Node, React, Java) with failing tests to fix using an AI assistant. The work-style survey has agree-disagree statements and email scenarios: answer against the Leadership Principles and stay consistent, because questions repeat.
OA questions candidates reported:
| Question (our summary) | What it tests |
|---|---|
| Most money in exactly k consecutive bags, given non-overlapping ranges up to 1e9 | Coordinate compression, sliding window |
| Fix a car rental or movie app: wrong cars shown, new cars missing, edits lost | Debugging with AI |
| Fewest adjacent swaps to group 0s and 1s | Greedy counting |
| Jump between blocks to burn the most calories | Greedy ordering |
| Count subarrays with exactly k distinct values; longest bitonic subsequence | Sliding window, DP |
Amazon OA questions shared on Codinzhub:
| Question | Drive | Topic |
|---|---|---|
| Maximum Consecutive Sum | ISM Dhanbad | DP |
| Maximize the Bottom-Half Warehouse Loads | IIT BHU | Greedy |
| Two Sum | IIT BHU | Hashing |
| Maximize Zero-Sum Factor | IIT Patna | Prefix sums |
| Maximum Continuous Uptime | LNMIIT Jaipur | Sliding window |
| Minimum Time to Fulfill All Orders Using Conveyor Belts | MNNIT | Binary search |
| Minimum Capacity to Ship Packages Within D Days | NIT Bhopal | Binary search |
| Find Minimum Sum | HackOn | Arrays |
| Secure Maximum Deliveries | HackOn | Greedy |
| Amazon Throttling Mechanism | Not stated | Simulation |
| Number of Unique XOR Triplets | Not stated | Bits |
| MovieDB Actor Follow Notifications (AI debugging) | Not stated | Web debugging |
More questions: Browse Amazon questions.
Stage 4: Phone screen (mostly experienced hires)
At a glance
- Length: 45 to 60 minutes
- Questions per round: 2 or 3 LP questions and 1 coding problem
- What was asked: Ownership and Dive Deep stories, then a medium problem
- Most common: valid palindrome; dependency installation order
Questions candidates reported:
| Question (our summary) | Follow-up reported |
|---|---|
| Install a package after all its dependencies | Detect cyclic dependencies |
| Valid Palindrome | Edge cases and complexity |
| Clock-hand angle; find duplicates with unique results | None |
| Group products into categories from linked pairs | Transitivity |
More questions: Browse Amazon questions.
Stage 5: The interview loop
How it works. Three to five one-hour rounds, usually 15 to 20 minutes of LP questions, 30 to 35 minutes of coding, 5 minutes for your questions. The hiring manager runs one round and the Bar Raiser, a manager from another team, runs another. You may not know which round is which in advance.
Coding and problem-solving rounds
At a glance
- Length: about an hour, with LP questions first
- Questions per round: 1 or 2 problems, plus LP questions
- What was asked: heaps, stacks, graphs and topological sort, sliding windows, LRU caches
- Most common: LRU cache; course schedule style dependencies; top K frequent
Questions candidates reported:
| Question (our summary) | Follow-up reported |
|---|---|
| Top K frequent elements | The data arrives as a stream |
| Asteroid collision | Test your own edge cases |
| Course Schedule, or a package build order | Return the order; detect cycles with two sets |
| Sliding window with put and getMax for the last k values | O(k) space; monotonic deque |
| LRU cache | Write the linked list yourself |
| Merge three sorted arrays without duplicates | k arrays with a heap |
| Find all anagrams of a word in a stream | Fixed-size window |
| Meeting Rooms variant; longest substring without repeats | None |
More questions: Browse Amazon questions.
Low-level (object-oriented) design
At a glance
- Length: 30 to 45 minutes inside a round
- Questions per round: 1 design with code
- What was asked: pricing engines, ticketing, parking lots, file search
- Most common: a price calculator with changing rules; a parking lot
Questions candidates reported:
| Question (our summary) | Follow-up reported |
|---|---|
| Checkout price engine with percent, fixed and buy-X-get-Y discounts | Where each discount is applied |
| Train fare between stations | Add stations; cap weekend fares |
| Event ticket system with different ticket types | New pricing rules per type |
| A Unix find command that searches files by condition | Recursive search, extensible filters |
| Parking lot | Concurrency |
| Pizza ordering; simple pub/sub with priorities | Extensibility |
More questions: Browse Amazon questions.
System design (SDE II and above)
At a glance
- Length: about 45 minutes plus LP questions
- Questions per round: 1 design
- What was asked: data models, APIs, transactions, geospatial search
- Most common: a library or locker system; a ride-hailing app
Questions candidates reported: a library system across branches (data model, search API, the return transaction, no overlapping loans); a ride-hailing app with frequent GPS writes; Amazon Locker; a news feed; a product recommendation system.
LP and Bar Raiser rounds
At a glance
- Length: a full hour in the Bar Raiser round; 15 to 20 minutes in others
- Questions per round: 2 to 6 LP questions with many follow-ups
- What was asked: every Leadership Principle, with deep follow-ups
- Most common: Ownership, Dive Deep, Disagree and Commit, a missed deadline
Questions candidates reported:
- Something you owned that was not your responsibility.
- A production issue you dug into: how you confirmed the root cause.
- A time you disagreed with your manager, and a time you disagreed but committed.
- A tight deadline you met, and one you missed.
- Pushing back on a decision that hurt your team.
- A system failing intermittently while the obvious metrics looked fine.
- How you use generative AI, and how you check its output.
More questions: Browse Amazon questions.
Stage 6: Decision and offer
How it works. Decisions usually came within five business days; one offer came three days after the loop. One candidate passed but the role was filled, and was put on hold for a new match. One candidate with 1.5 years of experience was given an SDE I loop.
Amazon experiences shared on Codinzhub
| College | Role and result | What was asked |
|---|---|---|
| IIIT Allahabad | 6-month intern + PPO, selected | DSA round; a resume round where the interviewer checked the GitHub projects |
| DTU | SDE intern, selected | Burning tree; minimum platforms; House Robber III; topological sort |
| DTU (off campus) | 6-month intern, selected | Longest zero-sum subarray; cousin-sum tree update; frequency filter with a trie |
| ISM Dhanbad | SDE intern, not selected | Modified LRU cache; top K from a stream; tree DFS; C++ questions |
| Not stated | SDE 2, selected | Zigzag traversal; e-commerce LLD; Amazon Locker HLD; Course Schedule II |
| Not stated | SDE 1, selected after Bar Raiser | Rotting oranges; parse a Lisp expression; Best Time to Buy and Sell Stock III |
The Codinzhub Next-Round Playbook
What Amazon candidates who moved on did. Put together by the Codinzhub team from the Amazon experiences in this guide.
To get past the OA
- Practise one hard sliding-window or DP problem in 40 minutes, with large constraints.
- For the AI debugging part, read the failing tests first, then trace the route, view and query layers.
- Answer the work-style survey against the Leadership Principles, consistently.
To get past the loop
- Write 10 to 12 STAR stories and tag each with the principles it shows. Use each story once per loop.
- Expect follow-ups until "what would you do differently now"; have the technical detail ready.
- In coding, go brute force, then optimal, then test your own edge cases out loud.
- For low-level design, practise the common problems and ask how rules may change.
To get past the Bar Raiser
- For Disagree and Commit, focus on what you did after committing.
- Name a real weakness with a concrete improvement.
FAQ
Is there system design for new grads?
Usually a low-level design, not a full system design. SDE II and above get both.
How important are the Leadership Principles?
Very. Candidates said the LP part decided several outcomes, and one recruiter said a round would focus on two principles.
About this guide
This guide is researched and written by the Codinzhub team. We read every Amazon interview experience we could find, kept only what candidates actually described, and rewrote it in our own words. Questions are our own short summaries; the practice links point to similar problems, not the exact interview questions. Found something out of date? Share your own experience on Codinzhub and we will update the guide.
Keep track of what you’ve learned.