Talentee
Amazonadvanced

Amazon Software Engineer (SDE) Interview Questions

Amazon's interview is uniquely built around its 16 Leadership Principles. Every behavioral answer must map to one. The Bar Raiser holds veto power and is your hardest gate.

Process length
6-10 weeks
Rounds
6
Questions
10
Mid-level TC
$220k–$300k (SDE II)
Practice Amazon questions with AI

The Amazon Software Engineer (SDE) interview process

What to expect, in order.

  1. 1Online assessment (90 min — 2 coding problems + work simulation)
  2. 2Phone screen (45 min — 1 coding + 1 Leadership Principle question)
  3. 3Onsite round 1 — coding (60 min)
  4. 4Onsite round 2 — coding + design (60 min)
  5. 5Onsite round 3 — system design (60 min, for SDE II+)
  6. 6Onsite round 4 — Bar Raiser (60 min, behavioral-heavy with 1-2 LPs probed deeply)

What Amazon actually evaluates

Amazon is famously frugal and ownership-heavy. Engineers carry pagers, on-call rotations are strict, and tenure is short by industry standards (median ~1.5 years). The interview reflects this: every question maps to an LP, and the Bar Raiser will dig 4-5 layers deep on a single example to find inconsistencies.

Customer Obsession (the LP that trumps all others)
Ownership — long-term thinking, no 'not my job'
Invent and Simplify
Are Right, A Lot — judgment quality
Bias for Action — calculated risk-taking
Deliver Results
Insist on the Highest Standards

Process quirks worth knowing

The Bar Raiser is the unique Amazon mechanism: a trained interviewer from another team whose explicit job is to maintain hiring standards. They have veto power. Bar Raiser questions follow STAR + LP mapping rigorously — they will not accept vague answers and will keep drilling until they hit your real example or find the gap.

10 questions Amazon actually asks

Each question includes the tip for answering and what the interviewer is actually evaluating.

Q1behavioral

Tell me about a time you took a calculated risk that didn't work out.

Why Amazon asks: Maps to Bias for Action + Are Right A Lot. Bar Raiser wants to see you took ownership of the failure and extracted a real lesson, not blamed circumstances.
How to answer: Use STAR strictly. Pick a real example with quantified downside. Discuss what data you had, what you'd do differently, and how you applied the lesson to a subsequent decision.
What they evaluate: Genuine ownership of the failure, structured reflection (not just 'I learned to be careful'), evidence the lesson translated into action
Q2behavioral

Describe a time you disagreed with your manager. How did you handle it?

Why Amazon asks: Maps to Have Backbone; Disagree and Commit. Amazon wants engineers who push back hard with data, lose gracefully if outvoted, then commit fully to the chosen path.
How to answer: Show the disagreement was based on data, you presented it directly (not via passive aggression), and once a decision was made you committed without dragging your feet. Include the outcome.
What they evaluate: Comfort with conflict, data-driven argument, full commitment after the decision
Q3technical

Given a list of intervals, merge all overlapping intervals.

How to answer: Sort by start time. Iterate and merge: if current interval's start ≤ previous interval's end, extend the previous interval's end to max(prev.end, current.end). Otherwise push current as new.
What they evaluate: Recognition that sorting unlocks the linear pass, careful edge case handling (full containment, exact overlap), clean implementation
Q4behavioral

Tell me about a time you went above and beyond for a customer.

Why Amazon asks: Customer Obsession is THE LP. Every interview has at least one CO question. They want to see you actively chose the customer's interest at personal or team cost.
How to answer: Pick an internal or external customer story. Show you actively chose the customer's outcome over convenience. Quantify the customer benefit (saved hours, prevented incident, etc).
What they evaluate: Genuine prioritization of customer outcome over self-interest, evidence the customer noticed, scalable behavior (not a one-off)
Q5design

Design Amazon's product recommendation system.

Why Amazon asks: Amazon interviewers will probe both ML serving (latency, recommendation freshness) and business logic (cold start for new users, A/B test infrastructure).
How to answer: Start with two paths: candidate generation (collaborative filtering, content-based, popular) then ranking (ML model). Discuss real-time vs precomputed tradeoffs, A/B testing framework, and cold-start strategies.
What they evaluate: Understanding of two-stage recommender pattern, awareness of recommendation system cold start, ability to discuss at Amazon scale (millions of products, hundreds of millions of users)
Q6technical

Find the kth largest element in an unsorted array.

How to answer: Min-heap of size k: iterate the array, push to heap, pop if size exceeds k. The heap root is your answer. O(N log k). Or quickselect for O(N) average.
What they evaluate: Two valid approaches (heap and quickselect), ability to discuss tradeoffs, clean implementation of whichever you choose
Q7behavioral

Tell me about a time you said no to a request that wasn't right for the business.

Why Amazon asks: Maps to Have Backbone + Are Right A Lot + Customer Obsession. They want to see you can refuse requests that don't serve customers, even from leadership.
How to answer: Show the request, why it didn't serve the customer, how you proposed an alternative, and the outcome. Don't pretend you 'won' if you didn't — they value honest reporting.
What they evaluate: Willingness to push back, ability to frame the no in terms of customer impact, professional handling of disagreement
Q8technical

How would you optimize a SQL query that's running slow on a 1 billion row table?

How to answer: Walk through: EXPLAIN plan analysis, index strategy (composite vs single-column), query rewriting (subquery to JOIN, eliminating SELECT *), partitioning if dates are involved, caching at app layer.
What they evaluate: Knowledge of EXPLAIN, indexing fundamentals, awareness that optimization is empirical (measure first), partitioning at scale
Q9behavioral

Tell me about a time you had to make a decision with incomplete information.

Why Amazon asks: Maps to Are Right A Lot + Bias for Action. Amazon engineers operate with ambiguity constantly — they want to see you can make calibrated bets.
How to answer: Show: the decision, the gaps in your information, how you reasoned through the tradeoff (what data would I need to flip my decision?), and what happened.
What they evaluate: Calibrated reasoning under uncertainty, willingness to commit, post-decision learning
Q10values

Why Amazon specifically?

How to answer: Connect to a specific LP that resonates with how you work (Ownership, Bias for Action are common). Mention a specific team, product, or initiative. Avoid generic 'I want to work at scale'.
What they evaluate: Genuine alignment with at least one LP, specific knowledge of the team/role, signal of multi-year intent

Common ways candidates fail this interview

Specific to Amazon, not generic interview advice.

  • ⚠️Behavioral answers that don't map to an LP — the Bar Raiser will keep probing until you do
  • ⚠️Vague stories without specific numbers, dates, and outcomes
  • ⚠️Saying 'we did X' instead of 'I did X' — Amazon wants individual ownership
  • ⚠️Underweighting Customer Obsession — it overrides every other LP when they conflict
  • ⚠️Generic 'why Amazon' — they hear 'I want to work at FAANG' and reject you for it

Amazon Software Engineer (SDE) compensation (2026)

Entry / Junior
$165k–$200k total comp (SDE I)
Mid-level
$220k–$300k total comp (SDE II)
Senior+
$340k–$500k total comp (SDE III)

Sources: levels.fyi, Glassdoor, public filings (US figures, total compensation including base + bonus + equity).

Practice these questions with a live AI interviewer

Nova is Talentee's voice AI interviewer. Speak your answer out loud, get scored on structure, clarity, and confidence, with a detailed PDF report.

Start Free PracticeAll Software Engineer (SDE) questions