Amazon Interview Questions & Guide 2026
Amazon's interview is half algorithms, half Leadership Principles. 16 LPs evaluated across every round with Customer Obsession at the top, a Bar Raiser with formal veto power, no LeetCode hard problems, no dynamic programming, and no code execution. L4–L7 comp ($185K–$613K) with the infamous 5/15/40/40 Y1/Y2 cliff. Fastest FAANG timeline (24–32 days).
What makes Amazon different
Amazon's loop inverts the FAANG pattern. The recurring observation across interviewing.io and Exponent retrospectives is that strong coders who under-prepare Leadership Principles stories get rejected, while candidates with clean-but-not-brilliant code who carry the behavioral half get pushed over the line. Per interviewing.io: “If you have a bad LP round…it's almost always a no hire,” even if your coding was strong. That is not a decorative quote. It is the single most important sentence in this entire page. LP fit is the bottleneck. The coding bar is not.
The Bar Raiser is Amazon's second distinctive, and interviewing.io plus Exponent both describe it as the single round most likely to swing a FAANG outcome. A senior, specially-trained interviewer from outside the hiring team with formal veto power alongside the hiring manager. Their explicit job is to prevent the bar from being lowered to meet headcount pressure. Amazon has scaled back Bar Raiser rounds for some L4 roles in 2025–2026, but they remain standard for L5/SDE II and above. Uber has a Bar Raiser too, but Uber's version is technical-depth-heavy — Amazon's is LP-depth-heavy.
Coding is deliberately moderate. Amazon explicitly avoids LeetCode hard problems and rarely asks dynamic programming (Exponent + interviewing.io). The distribution is roughly 27% easy, 55% medium, 18% hard. Arrays, hash maps, trees, and graph traversal dominate. No code execution: Amazon's internal Livecode tool never compiles or runs your code. Jeff Atwood flagged the underlying signal years ago in Why Can't Programmers Program? — many candidates cannot talk through what they are writing while they write it, and that is exactly the skill Amazon's no-compile environment surfaces. The thinking is the signal. The pressure comes from the time-split with LP behavioral inside each 55-minute round, not from problem difficulty.
And then there's the comp cliff. Amazon's 5/15/40/40 RSU vest is heavily backloaded. Years 1 and 2 are covered by signing bonuses that expire at the end of Year 2. Gergely Orosz's Pragmatic Engineer coverage of FAANG compensation, combined with Levels.fyi submission data, makes the correct analysis explicit: when comparing an Amazon offer to Meta/Google/Apple, compute Year 3 and Year 4 TC separately, not just Year 1. An L5 offer quoted at $271K median only materializes in Years 3–4 and only if the stock tracks its grant-date price. If Amazon stock has not appreciated substantially between offer and Year 3, total compensation drops sharply at the Y2/Y3 boundary.
The interview loop
5-8 rounds: resume screen, recruiter call or 60-min HackerRank OA (seniors may skip), 30-40 min Livecode phone screen, then virtual onsite with 2 coding rounds, 1 dedicated LP round, 1 Bar Raiser, and system design at L5+. LP questions appear in EVERY round, not just the dedicated one.
Resume Screen + Recruiter Call
30–40 min · PhoneRole fit, timeline, location, level expectations, and light Leadership Principles warm-ups. Your recruiter will usually preview which LPs to prepare for. Senior engineers may skip the HackerRank OA by request; new grads cannot.
HackerRank Online Assessment
60 min · Automated / HackerRankgateAlgorithmic problems with automated scoring. Medium-leaning. The first scored gate for new grads. Amazon avoids LeetCode hard here — expect arrays, hash maps, trees, and graph traversal. No system design. Seniors may skip this round.
Technical Phone Screen (Livecode)
30–40 min · Live CodinggateAmazon’s internal “Livecode” tool with NO code execution — your code never compiles or runs. One coding problem plus 1–2 LP behavioral questions in the same round. Narrate your approach out loud; syntax errors and minor edge-case misses are forgiven, unclear thought process is not.
Onsite: Coding Round 1
55 min · Live Coding + LPgateOne or two medium problems (arrays, hash maps, trees, graph BFS/DFS). Roughly 50/50 split with Leadership Principles. Interviewer submits LP-tagged evidence against whichever principles the round is assigned to cover. Still no code execution in most loops.
Onsite: Coding Round 2
55 min · Live Coding + LPgateSecond coding round, usually a different pattern (heap / priority queue / topological sort / Top-K). LP questions continue — typically tagged to Dive Deep, Bias for Action, or Insist on the Highest Standards. Amazon rarely asks dynamic programming.
Onsite: Leadership Principles Deep-Dive
55 min · BehavioralgateThe dedicated LP round. 3–5 behavioral questions, each mapped to specific Leadership Principles. Expect 5–7 minutes of follow-up per story — the interviewer will probe for metric, trade-off, mistake, risk, and the mechanism you added afterward. This is the round where a weak story pool becomes obvious.
Onsite: Bar Raiser
55–60 min · Cross-team InterviewergateSenior, specially-trained interviewer from outside the hiring team with formal veto power alongside the hiring manager. May be present or scaled back for L4 in 2025–2026, standard for L5+. Probes LP depth, honest failure narratives, and long-term thinking. Their job is to prevent the bar from being lowered for headcount.
Onsite: System Design (L5+ only)
55–60 min · Whiteboard / VirtualgateSkipped at L4 — no system design for new grads. At L5+ expect Amazon/AWS-flavored questions: CloudFront, ElastiCache, product catalog typeahead, rate limiter, logistics routing, warehouse automation. Amazon prioritizes cost-aware, fault-tolerant reasoning over pure scalability diagrams.
Amazon behavioral questions, LP-tagged
Every Amazon behavioral question maps to 1–3 Leadership Principles. These are the most frequently reported LP-tagged questions across the SWE loop. Each answer should be a real engineering story from your story bank, not a script rewritten per question.
- 1
“Tell me about a time you went above and beyond for a customer.”
Maps to: Customer Obsession, Ownership
- 2
“Describe a time you had to make a decision with incomplete information.”
Maps to: Bias for Action, Are Right A Lot
- 3
“Tell me about a time you disagreed with your manager or a peer. What did you do?”
Maps to: Have Backbone; Disagree and Commit, Earn Trust
- 4
“Describe the most complex technical problem you’ve solved. Why was it hard?”
Maps to: Dive Deep, Deliver Results
- 5
“Tell me about a time you took on something significantly outside your role.”
Maps to: Ownership, Learn and Be Curious
- 6
“Describe a time you refused to compromise on quality under deadline pressure.”
Maps to: Insist on the Highest Standards, Deliver Results
- 7
“Tell me about a time you made a mistake that affected customers. How did you recover?”
Maps to: Customer Obsession, Earn Trust, Ownership
- 8
“Describe a time you had to simplify a complex system or process.”
Maps to: Invent and Simplify, Frugality
- 9
“Tell me about a time you had to influence a decision without direct authority.”
Maps to: Earn Trust, Have Backbone; Disagree and Commit
- 10
“Describe a time you proposed something bold that your team initially resisted.”
Maps to: Think Big, Have Backbone; Disagree and Commit
The Bar Raiser — what you actually need to know
Senior, specially-trained interviewer from outside the hiring team. Formal veto power alongside the hiring manager. Their job is to prevent the bar from being lowered for headcount pressure — they ask whether this hire raises the long-term bar or just fills a seat. Unlike Uber's Bar Raiser (technical depth), Amazon's Bar Raiser is LP-depth-heavy. Achieving Bar Raiser status is itself a promotion signal toward L6/L7.
How to prepare: Pick one flagship project with real numbers, then strip it down to five elements the Bar Raiser will probe for: metric (baseline and after), trade-off (what got worse or was postponed), mistake (what you got wrong the first time), risk (what you knowingly accepted), and mechanism (what you added afterward so it would not repeat). Stories without these elements fail this round.
Answer shape: Problem → Ownership → Decision → Trade-off → Result → Lesson. Weakest signal: saying “we” when the interviewer needs “I,” staying at team-level impact, hiding the trade-off. Strongest signal: named mistakes, honest numbers, scope another team would trust.
Difficulty breakdown
Amazon explicitly avoids LeetCode hard problems and rarely asks dynamic programming. Arrays, hash maps, trees, and graph traversal dominate. Difficulty pressure comes from the LP time-split inside each 55-min round, not from problem complexity.
Unlock the full guide
Complete walkthrough, diagrams, and practice problems — all included with StrongYes Pro.
Unlock with ProThe 16 Leadership Principles — SWE loop relevance
Amazon evaluates every engineer against all 16 Leadership Principles. Do not write 16 separate STAR scripts. Build a story bank of 6–8 real engineering stories, each tagged to 2–4 LPs, and rehearse the tagging. These are the LPs that show up most in SWE loops:
- Customer Obsession — Did you change the plan because a real user was getting hurt? The preeminent LP — when in doubt, route here.
- Ownership — Did you solve the problem even when part of it sat outside your scope? Never say “that’s not my job.”
- Dive Deep — Did you go below dashboards to find the real failure mode? Bar Raisers love metric-vs-anecdote contradictions.
- Bias for Action — Could you move with incomplete info without being reckless? Calculated risk-taking on reversible decisions.
- Earn Trust — Could you handle disagreement without becoming defensive? Be vocally self-critical about your own mistakes.
- Insist on the Highest Standards — Did you refuse to ship something half-correct? Raise the bar for the team, even when it is unpopular.
- Have Backbone; Disagree and Commit — Could you push back clearly, then fully commit once the decision was made? Both halves matter equally.
- Deliver Results — When deadline pressure was real, did the outcome actually improve? Outcome quality, not just shipping.
Senior and principal loops (L6/L7) additionally probe: Think Big, Hire and Develop the Best, Strive to be Earth’s Best Employer, Success and Scale Bring Broad Responsibility. At this level expect LP questions about hiring decisions, long-horizon bets, org-level trade-offs, and broader responsibility beyond immediate deliverables.
The remaining four LPs round out the 16. They come up less often in SWE loops than the eight above, but interviewers will still ask about them — especially when your resume or a story gives them a natural hook:
- Invent and Simplify — simplifying a gnarly system, replacing a vendor with an in-house primitive, or cutting scope to ship.
- Are Right, A Lot — a call you made that looked wrong and turned out right (or vice-versa) and what your decision process actually was.
- Learn and Be Curious — a technology or domain you picked up under deadline and what you did once you were the expert.
- Frugality — cutting cost or resource use without degrading the product. Strongest when the story is about AWS bill or infra spend.
Canonical source: Amazon's official Leadership Principles page. The two newest (added July 2021): Strive to be Earth's Best Employer and Success and Scale Bring Broad Responsibility.
New grad entry (L4 / SDE I)
New grads enter at L4 (SDE I) with median TC ~$185K ($136K base + $37.1K/yr stock + $11.2K bonus). Amazon is one of the highest-volume new-grad hirers; interns report 74% positive experience (highest in the FAANG store at entry level). There is no L3 at Amazon — L4 is the entry level.
What's different for new grads:
- No system design at L4. System design starts at L5/SDE II. Advantage vs Google L3 where LLD is increasingly common.
- HackerRank OA is unskippable. 60-min automated assessment, medium-leaning problems, no LeetCode hard. Seniors may skip this round by request; new grads cannot.
- LP bar is lower for entry level — which is why intern satisfaction is 74% positive. But LP failure is still terminal. Build 4–6 real stories, ideally from internships or significant class projects, each tagged to 2–3 LPs.
- 32-day timeline for SDE I Intern (Glassdoor average). Fastest FAANG loop for entry-level.
- Multi-team interviewing is permitted. Unique among FAANG. Apply broadly — if one team rejects you, another can still hire you.
- Customer Obsession is the preeminent LP. When in doubt, route your answer to how it helped a real user.
- Y1/Y2 cliff warning: 5/15/40/40 vest means your quoted TC only materializes in Years 3–4. Signing bonuses cover Y1/Y2 and then expire. Compare offers on Year 3 TC, not Year 1.
- Expected progression: L4 → L5 in 2–3 years. L5 → L6 in 3–5 years. L5 is “terminal level” — most Amazon engineers plateau here.
Interview culture
Candidates consistently describe Amazon interviews as the most behaviorally demanding of the FAANG group. The technical bar is below Google and Meta (Amazon avoids hard problems and rarely tests DP), but the Leadership Principles bar is higher and more consequential — LP performance can determine hire/no-hire independent of coding. Glassdoor data backs this up: Amazon SWE rates 48% positive (vs Meta 57%, Google SWE 62%) with 3.3/5 difficulty.
Negative sentiment concentrates on two things: the behavioral round being exhausting because LP questions repeat across every round, and the Bar Raiser adding unpredictable pressure since you cannot know which interviewer they are. Positive sentiment concentrates on speed — Amazon's loop is the fastest FAANG loop by Glassdoor timeline data (24–32 day averages), and Amazon is known for fast offer turnaround once you pass debrief.
Interns report the most favorable experience (74% positive) because the LP bar is lower for entry-level candidates. If you can intern at Amazon first, the full-time conversion path is easier. Customer Obsession is described by multiple sources as the “preeminent” LP — when in doubt, route your answer there.
Amazon's process is also the only FAANG to permit simultaneous multi-team interviewing. Each team runs its own loop. If one team rejects you, another can still hire you. This is a significant advantage for new grads applying broadly and for candidates who failed a previous Amazon loop on team fit rather than bar. AI tools are strictly prohibited during interviews, consistent with the 2026 FAANG tightening.
Offer strategy — reading an Amazon package
When Amazon quotes you $271K TC at L5 or $185K TC at L4, ask which year. The 5/15/40/40 RSU vest means only 5% of your equity vests in Year 1, 15% in Year 2, 40% in Year 3, 40% in Year 4. Signing bonuses — split across Year 1 and Year 2 — cover the early-year gap. Then they expire.
Practical implication: Compare Year 3 TC, not Year 1 TC. An Amazon L5 offer at $271K median only hits that number in Years 3–4 and only if the stock tracks its grant-date price. A Meta E5 offer at $420K hits $420K every year. If Amazon stock drops 30% between offer date and Year 3, your L5 TC can realistically drop to $200K–$220K in Year 3 even if all the grants vested.
The bonus column on Levels.fyi is deceptive too. L4 shows $11.2K, L5 shows $3.6K, L6 shows $1.1K, L7 shows $0. This is not because senior engineers get smaller bonuses — it's because “bonus” is primarily the signing-bonus amortization, and senior engineers have more equity and less reliance on signing bonuses. The real performance-bonus component at Amazon is small across all levels.
Amazon does negotiate, especially for candidates with competing offers. Base is close to locked at each level, but equity and signing bonuses have more flex. Use a competing Meta/Google offer to push equity. Use a Year 3 TC calculation to push the signing-bonus split (front-loading more to Year 1 is worse for you long-term than getting more equity).