Amazon runs one standardised hiring machine across wildly different job families: an online assessment or recruiter screen, a phone screen, then a 'loop' of roughly four to six one-hour interviews with individual employees, one of whom is a Bar Raiser drawn from outside the hiring team and widely described as the hardest hour. What changes between families is only the technical half of each hour: coding plus low-level and high-level design for engineers, science breadth/depth plus a tech talk for scientists, AWS and Linux troubleshooting for cloud support, SQL for analysts, a timed Excel or written case for finance, and nothing technical at all for warehouse operations. The constant is the behavioural half, where each interviewer is assigned specific Leadership Principles and expects STAR-shaped stories with hard numbers; candidates across every family report that this, not the technical work, is what decided their outcome.
These 26 writeups cover software engineering roles, India.
Typical rounds
3.41-7 range
Outcomes shared
14/9offer / not
Most common round
Technical
Sources span
2014 - 2026
Most frequently reported · Leadership Principles(17), Work simulation(3), Data structures & algorithms(2), STAR stories(2), Work style survey(2)
Amazon’s official process
Amazon's own How We Hire hub breaks the process into four named stages: Online Application, Assessments, Phone Screening and Interview Loop, and states up front that the shape differs by role. It describes the loop as meeting current employees individually, with each person assessing a different aspect of your skills and experience, and its published loop-prep material is organised around the Leadership Principles, behavioural-based questions and the STAR method rather than around technical topics. The site carries separate role-specific prep pages not only for engineering (SDE, SDE II, SDE III, SDM, front-end, BIE, security) and science roles, but explicitly for non-technical ones (customer service manager, legal, marketing manager, product manager, program manager, vendor manager) and for university hiring, including a Field Manager Assessment used for operations roles. It also advertises optional Candid Chats with employee-led affinity groups during the process, and accommodations on request.
23,214 open roles in our index · advertised pay floors run $113k-$158k (median $137k, from 230 postings that state pay).
Most-posted roles
Area Manager II 286
Data Center Technician222
Operations Manager209
Engineering Operation Technician152
Area Manager119
Sr. Operations Manager119
Top locations
Seattle3386
New York City1225
Bengaluru1107
Bellevue1106
Hyderabad881
Commonly tested topics
Leadership Principles, assigned per interviewer and probed in nearly every round·STAR-format behavioural storytelling with concrete metrics and aggressive follow-ups·The Bar Raiser: an interviewer from outside the hiring team, usually the hardest round·Online assessment combining coding, a work simulation and a work-style consistency survey·Data structures and algorithms: arrays, strings, sliding window, stacks, trees, graphs, heaps, DP·Low-level / object-oriented design of a small system, with working code expected inside the hour·High-level system design including failure modes, retries, fallbacks and product trade-offs·Science breadth, science depth and a presented tech talk for applied/research scientist roles·AWS services, networking (OSI, TCP, DNS, SSL), Linux and OS troubleshooting for cloud support and TAM·SQL and Excel for analyst roles; time-boxed Excel or written cases for finance roles·Written exercises: Amazon's writing culture surfaces as take-home or in-person writing prompts·Time management inside the round: several candidates lost on failing to start coding early enough·Down-levelling (L6 to L5, L5 to L4) offered instead of a clean reject·Long, opaque post-loop waits and a firm no-feedback policy on rejection·Data structures & algorithms·Leadership Principles·Behavioral (STAR)·Bar Raiser·Online assessment
Interview experiences at Amazon
Outcome
Seniority
Software Development Engineer II (L5)
Experienced
Not selected5 rounds
1
Phone screenPhone screen
A single algorithmic problem about counting contiguous subarrays summing to a target, solved with a running-sum-plus-hash-map approach. The candidate handled the follow-ups comfortably and rated this his strongest round.
Bar Raiser (low-level design + Leadership Principles)Bar raiser
The Bar Raiser came first rather than last and unexpectedly centred on low-level design: after about fifteen minutes on background and Leadership Principles, he was asked to model a multi-stage audio-buffer pipeline. He drifted into high-level architecture, was asked to code with roughly twelve minutes left, and could not finish.
Ten minutes of Leadership Principles, then a file-system cache design using an LRU policy backed by a linked list and a hash map. Discussion covered whether to move or recreate an accessed node and whether the cache should hold metadata or full file contents; he ran out of time before finishing every method.
A high-level design of a third-party service pushing a feature announcement to users through a voice assistant. After being redirected away from entity modelling toward the end-to-end flow, he covered the happy path then failure handling for offline devices, and closed with a short discussion of how he uses generative AI in his workflow.
Covered · service decomposition, notification delivery, offline devices, retries and fault tolerance, product trade-offs
5
Coding (graph)Technical
A short Leadership Principles and resume section followed by a dependency-ordering graph problem.
Two coding problems of moderate and high difficulty, followed by Amazon's work-simulation and work-style components. He solved one fully and one partially and still advanced.
Covered · two algorithmic problems, work simulation, work style assessment
2
Coding interviewTechnical
Seventy minutes covering one medium and one hard problem, with feedback afterwards that his problem solving was fine but his explanation of past work was weak.
Covered · greedy / jump-game style problems, binary tree covering
3
Coding plus Leadership PrinciplesTechnical
Ninety minutes split between two algorithm problems and four behavioural questions. The interviewer gave no live feedback beyond confirming his communication was fine.
Covered · stacks / greedy digit removal, shortest path with cost and time constraints, Leadership Principles: pride in work, diving deep, independent delivery, communicating slippage
4
Bar RaiserBar raiser
A short thirty-five-minute round with a long-tenured engineering manager, spent almost entirely on one deep question about a complex problem he had researched and prototyped. He felt he had failed it; he got the offer two days later.
Covered · Dive Deep, Earn Trust, Customer Obsession, a complex research-driven problem from his own work
A hosted coding challenge with two problems plus scenario questions about how to prioritise items in a hypothetical project.
Covered · two medium-difficulty algorithm problems, work-prioritisation scenario questions
3
Virtual onsite, interview 1Technical
The recruiter had briefed him that all four loop rounds would be roughly half behavioural and half technical, and the first followed exactly that shape.
Covered · two Leadership Principle questions, medium algorithm problem
4
Virtual onsite, interview 2System design
Same split, with the technical half given over to a system design discussion.
Covered · two Leadership Principle questions, high-level system design
5
Virtual onsite, interview 3Technical
A second coding round on the following day, again preceded by two behavioural questions.
Covered · two Leadership Principle questions, medium algorithm problem
6
Virtual onsite, interview 4Low-level design
The final hour paired two more Leadership Principle questions with a low-level design problem, a format he had not realised existed until the day before.
Covered · two Leadership Principle questions, low-level / object-oriented design
A tree traversal problem and an implementation of a frequency-based cache eviction algorithm.
Covered · binary tree top view, cache replacement policy implementation
4
System designSystem design
A single open-ended design of a related-items recommendation engine, covering both architecture and class structure.
Covered · recommendation engine, high-level and low-level design
5
ManagerialPanel
A mixed round blending behavioural questions with detailed interrogation of his own systems, including being asked for exact numbers on performance work.
Covered · Leadership Principles, architecture of his current projects, design justification, quantified performance improvements
6
Hiring managerHiring manager
A video conversation focused on self-critical retrospectives: failed designs, production breakage he was responsible for, and proposals of his that shipped.
Covered · hardest problem solved, a design that failed, evaluating alternative systems for a use case, a production incident he caused
A four-section campus test covering debugging, two coding problems, an aptitude section and a personality/leadership questionnaire. Forty-one students from his cohort advanced.
A virtual interview that opened with roughly ten minutes on his own project, then two problems where he was pushed from brute force to optimal, asked to derive complexity formally, and asked to dry-run his own code on a test case he invented.
Covered · own project deep dive, binary search trees, tree subtree-sum property, sliding window with a flip budget, time and space complexity
A one-hour behavioural assessment in three parts. A senior already at Amazon told him this was the round that actually filtered people, because the coding half was easy. The instrument deliberately repeats similar items to test whether your answers stay consistent.
Covered · agree/disagree statement batteries, choosing the best reply to workplace emails, forced-choice personality items, answer consistency
3
Technical interviewTechnical
A single round on Amazon Chime with two panellists and two standard problems, worked from brute force through binary-search optimisation. He performed well but was cut because the actual headcount landed at eight rather than the roughly twenty candidates the cohort had been told to expect.
Covered · search in a sorted 2D matrix, symmetric binary tree
A second-round assessment administered by a third-party proctoring service that took remote control of his machine, disabled screenshots, required a full camera sweep of his room and desk, and barred pen, paper and any written material. The work-simulation video then failed to load; after roughly an hour of failed page refreshes he withdrew from the process mid-assessment.
Covered · remote proctoring and machine takeover, video-based work simulation, timed test conditions
A roughly five-hour social event the night before the onsite.
Covered · informal mixer
2
Onsite: four whiteboard roundsTechnical
Four forty-five-minute whiteboard coding sessions across a full day, two of which were conducted by two interviewers at once. He received no debrief and no explanation of which round he failed, contrary to the older published description of the process.
Rounds 1-8: full science loop over roughly two monthsPanel
An unusually long eight-round process stretched across about two months, one of which was a forty-five-minute tech talk followed by fifteen minutes of questions.
Received an offerMonthly Salary: $9500 (plus $6,000 relocation)1 round
1
Rounds 1-3: three dated interviews across two weeksTechnical
Three interviews on 11, 12 and 18 June 2019, with the offer landing on 25 June. The write-up is organised around the question areas he was asked rather than a formal round taxonomy.
Covered · statistics, machine learning, data manipulation, Leadership Principles
A recruiter screen in early March, roughly seven weeks after he applied in mid-January.
Covered · background, role fit, logistics
2
Hiring managerHiring manager
A one-hour conversation with the hiring manager in late March.
Covered · experience, Leadership Principles
3
Recruiter prep callPhone screen
A dedicated preparation call from the recruiter ahead of the virtual loop.
Covered · loop format, Leadership Principles briefing
4
Virtual loopPanel
Six or seven one-hour sessions on Amazon Chime in a single day, including a Bar Raiser, with each interviewer carrying their own assigned set of Leadership Principles. The offer came a week later.
Covered · AWS technical breadth, Leadership Principles (two or three assigned per interviewer), customer scenarios
A campus test with twenty aptitude questions, a core-CS section and thirty minutes of coding. Around six hundred people sat it; fifty-four were interviewed the same day.
A simulation scoring the quality of his decisions, which he understood to be continuously assessing whether he behaved like an Amazon leader.
Covered · decision quality, consistency with Leadership Principles
3
Interview seriesBehavioral
A sequence of behavioural interviews. He was run against two different roles in parallel; one went noticeably better and it was the Pathways one he actually wanted.
One hour split between two interviewers at thirty minutes each, entirely behavioural.
Covered · Leadership Principles, data analysis, handling a difficult colleague
2
Final interviewFinal
The same format again, two interviewers at thirty minutes each. Across both rounds she was asked roughly eight to twelve behavioural questions and no technical or case questions at all.
Covered · Leadership Principles, a mistake she made, disagreeing with management
An hour-long screen roughly twenty days after the recruiter's first contact, consisting entirely of situational behavioural questions.
Covered · three situational Leadership Principle questions
2
Writing assessmentTake-home
A written exercise where she picked one of two prompts and returned a STAR-structured response within about forty-eight hours, submitted two days before the onsite.
Covered · written STAR narrative, choice of two prompts
3
Onsite loopPanel
Five one-hour rounds spread over two days, all behavioural, including a Bar Raiser and some rounds with shadowing observers. She was told she was better suited to L5 and placed in a six-month talent pool rather than rejected outright.
Four in-person product interviews, almost all anchored on real Amazon products rather than abstract prompts, with the Bar Raiser the hardest of the set. He chose Marketplace over AWS and Kindle after the offer.
Received an offera "$25,000 pay cut", offset by anticipated four-year stock; she notes "Amazon does not really negotiate"2 rounds
1
Attempt 1 (2021): Onsite loopPanel
An eight-hour loop across two days with eight interviewers including a Bar Raiser. She was rejected with the feedback that she needed more technology-sector experience.
An entirely behavioural loop where each interviewer spent an hour probing his fit against the principles. By the third interview he had run out of distinct stories and was repeating himself.
The Bar Raiser was swapped at the last minute for someone who, in his account, did not understand the role and spent the hour interrupting with unrelated questions. He got the offer and declined it, citing the interview experience itself.
The Bar Raiser round, mixing further Leadership-Principles probing with a technical problem. This candidate's loop ended in a rejection at the final stage.
Covered · Leadership Principles, Data structures & algorithms
Amazon runs one standardised hiring machine across wildly different job families: an online assessment or recruiter screen, a phone screen, then a 'loop' of roughly four to six one-hour interviews with individual employees, one of whom is a Bar Raiser drawn from outside the hiring team and widely described as the hardest hour. What changes between families is only the technical half of each hour: coding plus low-level and high-level design for engineers, science breadth/depth plus a tech talk for scientists, AWS and Linux troubleshooting for cloud support, SQL for analysts, a timed Excel or written case for finance, and nothing technical at all for warehouse operations. The constant is the behavioural half, where each interviewer is assigned specific Leadership Principles and expects STAR-shaped stories with hard numbers; candidates across every family report that this, not the technical work, is what decided their outcome.
What topics does Amazon test in interviews?
Commonly reported topics include Leadership Principles, assigned per interviewer and probed in nearly every round, STAR-format behavioural storytelling with concrete metrics and aggressive follow-ups, The Bar Raiser: an interviewer from outside the hiring team, usually the hardest round, Online assessment combining coding, a work simulation and a work-style consistency survey, Data structures and algorithms: arrays, strings, sliding window, stacks, trees, graphs, heaps, DP, Low-level / object-oriented design of a small system, with working code expected inside the hour.
These guides summarize public, first-hand interview experiences shared by candidates. They describe the shape of each loop and the topics that came up, not a leaked question bank, and every experience links back to its original source. Processes change often; treat this as directional prep, not a script.
