Amazon Software Engineer Interview Guide: DSA, LPs, and the Bar Raiser

Most people prep for Amazon like they're preparing for a Google interview. Then they walk in and discover the coding problem is only half the evaluation. Every round runs two simultaneous assessments and the interviewer shows up with two clipboards.

The structural difference is this: Amazon runs both a technical interview and a behavioral interview inside every single round. Both get scored. Both go into the debrief. Prep for only one and you're already behind before you write a line of code.

The Full Loop at a Glance

Stage	Format	Duration	What's Assessed
Online Assessment	2-3 coding problems	70-90 min	DSA fundamentals
Phone Screen	1 DSA problem + 1 LP	45-60 min	Coding + behavioral fit
Onsite Loop	4-6 rounds	~5 hours total	DSA, system design (senior), LPs
Bar Raiser	1 round within loop	45-60 min	Judgment, depth, senior bar

Timeline from application to offer: four to eight weeks. Block your calendar accordingly.

Round 1: The Online Assessment

The OA is your first filter, and Amazon made it harder in 2025. They randomize constraints, rename problems, and shuffle inputs between candidates. Memorizing question dumps stopped working. The internet is full of people who tried.

You get two to three coding problems in roughly 80 minutes. Difficulty sits at LeetCode medium, occasionally leaning hard. There's also a Work Simulation and Work Style survey, but neither kills you the way a wrong answer in the coding section does.

The most common OA patterns: sliding window, graph traversal, string manipulation, and greedy. Write clean, correct solutions fast. No second shots here.

The Phone Screen: One Problem, Two Signals

Pass the OA and a recruiter schedules a technical screen with a team member. One DSA problem, one Leadership Principles question, 45 to 60 minutes.

The coding problem is standard medium difficulty. The LP question is usually something like "Tell me about a time you disagreed with your manager" or "Tell me about a project you owned end to end." This is not a warmup. Your answer goes into the hiring notes and gets read aloud in the debrief alongside your code.

A weak LP answer here has killed strong coding performances. Amazon uses this screen to decide whether to invest five hours of loop time in you. They are not being casual about it.

The Onsite Loop: Five Rounds, Each With Two Jobs

The loop is 4 to 6 rounds with different interviewers. Every interviewer has two assignments: one technical focus and one or two Leadership Principles to probe.

That means every round has a mode switch. Coding ends, behavioral begins. Both results land in the debrief.

Typical breakdown:

Coding rounds (2-3): One medium-to-hard DSA problem each. Arrays, trees, graphs, dynamic programming, heaps. One problem per round, 20-30 minutes to solve, the rest for discussion and LP.
System design (SDE 2 and above): One round on distributed systems, scalability, or service architecture.
Hiring manager round: More conversational, heavy LP, lighter technical.
Bar Raiser round: Coming up.

The Bar Raiser Has Veto Power. You Won't Know Who It Is.

Every loop includes one Bar Raiser: a trained Amazon employee from outside the hiring team, nominated by the recruiter. Their job is to maintain the hiring bar across the entire company, not just your team.

The Bar Raiser can veto a "no hire" even if every other interviewer wants to extend an offer. The reverse is not true. If the hiring manager says no, the Bar Raiser cannot override. Both have to agree to hire.

You cannot tell which round it is. The meeting looks like any other. Bar Raisers tend to go deeper on follow-ups and revisit LP answers from earlier in the loop, so if you said two different things about the same project in two different rounds, expect it to surface. Teams under hiring pressure will rationalize weak candidates. The Bar Raiser is the structural check against that.

Cat sitting at a counter with paws folded, maintaining composure like a professional interviewer The Bar Raiser watching you deliver three contradictory versions of your "Dive Deep" story across the loop

DSA: What Amazon Actually Asks

Amazon's DSA is weighted toward medium-level problems. Hard problems appear more frequently at SDE 2 and above, but the loop is not designed to break you with obscure tricks.

Common topic areas:

Arrays and strings (two pointers, sliding window, prefix sums)
Trees and graphs (BFS, DFS, traversal)
Dynamic programming (1D and 2D, memoization, bottom-up)
Heaps and priority queues (k-th element, merge, scheduling)
Hashmaps and hash sets (frequency counts, complement lookups)

Amazon interviewers care about your process, not just your answer. Narrate your approach before writing code. Handle edge cases explicitly. Explain your time and space complexity. A correct solution delivered in silence is a red flag, not a win.

See how to communicate during technical interviews for exactly how to narrate without sounding like you're reading from a teleprompter.

Leadership Principles: Amazon's Second Interview Inside the Interview

Amazon has 16 Leadership Principles. Every interviewer is assigned one or two to probe. Across the loop, most of the 16 will come up at least once. You thought you were signing up for a coding interview. You signed up for a coding interview plus a values audit.

The ones that appear almost universally:

Customer Obsession: What trade-offs did you make for the customer?
Ownership: What did you do that wasn't in your job description?
Dive Deep: Walk me through the technical detail of a problem you debugged.
Bias for Action: When did you make a decision without all the data?
Deliver Results: What did you actually ship?
Earn Trust / Vocally Self-Critical: Tell me about a mistake you made.

The STAR method is mandatory, but most candidates execute Action and Result badly. Action needs to be what you specifically did, not what the team did. "We built the pipeline" tells them nothing. "I redesigned the batching logic, which cut processing time by 40%" is what they need.

Result needs a number. "The project shipped on time" is not a result. "We reduced P95 latency from 840ms to 120ms and checkout conversion improved by 7%" is a result.

Prepare five to eight strong stories. Each should flex across multiple LPs. You can reuse stories across rounds, but interviewers compare notes in the debrief, so don't use the exact same example for every LP question in the loop.

Difficulty: Medium Problems, With a Communication Tax

The DSA ceiling at Amazon is lower than at Google or Jane Street. You are much less likely to see a novel hard problem requiring a specialized algorithm you've never encountered.

Amazon interview in a nutshell: the person who explains their Java solution clearly beats the silent genius every time

What you will face is the communication tax. Amazon explicitly evaluates how you talk through problems. A candidate who solves a medium correctly but never explains their reasoning scores lower than one who explains a slightly imperfect solution clearly and catches their own bug.

For SDE 1, two medium problems in the loop is normal. For SDE 2, expect one medium and one problem that requires real insight or optimization. Senior and above adds system design and heavier LP scrutiny.

Amazon Software Engineer Interview Prep: Where the Hours Should Go

If you have six to eight weeks:

DSA (50-60% of time): Practice medium problems across the topic areas above. The sliding window and dynamic programming patterns come up repeatedly. Practice narrating out loud while you solve. If you can't explain your approach before you code it, you are not ready.

Leadership Principles (30-35% of time): Write out your stories. Literally write them. Identify the LP each story demonstrates, find the quantified result, and practice delivering it in under two minutes. Run through all 16 and confirm you have at least one story per principle.

Mock interviews (10-15% of time): Practicing on paper and practicing out loud are different skills. The interview is a spoken performance. SpaceComplexity runs voice-based mock interviews with rubric-graded feedback on exactly the dimensions Amazon evaluates: communication, problem-solving, code quality, and behavioral depth.

Use clarifying questions strategically. Amazon interviewers expect you to drive the problem-understanding phase before touching code.

Mistakes That Kill Otherwise Good Candidates

Treating LP questions as a formality. This is the most common and most expensive mistake. Amazon cares more about behavioral signal than any other major tech company. Weak LP answers fail candidates who solve the DSA perfectly. Ask anyone who has debriefed at Amazon.

Using "we" throughout behavioral answers. Interviewers are assessing you. "We built the pipeline" tells them nothing. "I redesigned the batching logic, which cut processing time by 40%" is what they need.

Going silent during coding. Amazon interviewers evaluate communication as a separate signal. Five minutes of quiet while you think is a negative data point, even if you surface with a correct solution. Thinking is fine. Thinking visibly is better.

Inconsistent stories across rounds. The debrief is a group conversation. If you told two different versions of the same project story in different rounds, it will come up. Your LP answers need to be true and consistent, not just polished.

Not asking clarifying questions. Jumping straight into code signals that you skip problem understanding in real work too. One or two focused questions before you start shows you think before you build.

Ignoring edge cases. Amazon explicitly tests whether you handle them. Ask, then handle.

The Eight-Week Timeline

8 weeks out: Audit your DSA gaps. Run through the core topic areas and identify where you are slow or shaky. Start your LP story collection in a doc.

6 weeks out: Focused problem practice, 5-7 problems per week, timed, with verbal narration. Finalize your LP story bank.

4 weeks out: Full mock loops including behavioral. Dial in your STAR delivery. Practice consistency across LP answers.

2 weeks out: Review your weakest DSA areas, refresh LP stories, run two to three full mock interviews end to end.

1 week out: Light review only. No new material. See how not to waste your final week for exactly what to do and what to skip.

Key Takeaways

Amazon's loop is 4-6 rounds including OA, phone screen, onsite, and a Bar Raiser with veto power on no-hire.
Every round includes a DSA component and a behavioral component. Both are scored.
DSA difficulty peaks at medium-to-hard; communication quality is evaluated separately and explicitly.
16 Leadership Principles are divided across interviewers. You will face most of them.
The STAR method only works if your Action is specific and your Result is quantified.
The Bar Raiser is not a harder technical round. It is a depth check on judgment and consistency.
Prep split: roughly 55% DSA, 35% LP, 10% mock interviews.