How Amazon interviews — the bar raiser

1. The shape of an Amazon loop

Amazon runs the most documented interview process of any large tech company. They publish the Leadership Principles. They tell you the role of the bar raiser. They send you a prep doc. There is no hidden ritual here. The challenge is that the rubric is wider than other companies — they grade you on a mix of technical skill and behavioral fit, in every round, simultaneously.

A standard SDE2 (mid-senior) onsite is five rounds:

Round 1   Coding / problem solving             (60 min)
Round 2   System design                        (60 min)
Round 3   Coding deep dive on a past project   (60 min)
Round 4   "Pure" behavioral (LP-heavy)         (60 min)
Round 5   Bar raiser — mixed                   (60 min)

Five hours of interview. One day. The bar raiser is the most-talked-about and the most-misunderstood part of it.

2. The bar raiser, explained

The bar raiser is a senior engineer from a different team who has been trained and certified specifically to keep the hiring bar consistent across Amazon. They have one job: make sure this hire would raise the bar on the team they're joining, not lower it.

Things to know about the bar raiser:

• They have veto power. Even if the hiring manager wants you and every other round is positive, the bar raiser can reject. The reverse is also true — they can push through a hire other interviewers were lukewarm on.
• They are not on the hiring team. They have no incentive to like you. Their incentive is bar quality.
• They run a mixed round. Coding, design, behavioral — often all three in one hour. They pivot based on what they want to verify.
• They take the longest writing the feedback. Their feedback is the most-read by the hiring committee.

If you walk out of an Amazon loop and the bar raiser round felt the most awkward — that is normal and possibly a good sign. They are trying to break the candidate's comfort.

3. The Leadership Principles are the technical rubric

Most candidates treat the LPs as a behavioral matter — "I'll tell a Customer Obsession story in the behavioral round." That's half right. The LPs are graded in every round, including the technical ones.

A few that show up explicitly in system design rounds:

The strong candidate name-checks one or two of these without sounding like they're checking a box. "This is a Frugality call — we don't need a queue here, a direct write is fine."

4. The level bar

Amazon levels for software engineers:

• SDE1 (entry). Almost no system design. Coding-heavy. Light LP grading. Skip this course.
• SDE2 (mid-senior, the most-hired level). One system design round. Single deep dive is enough. LPs graded continuously but at a forgiving bar.
• SDE3 (senior). Two system design rounds, both deep. Multiple deep dives expected. LP bar tightens — Dive Deep and Are Right, a Lot are no longer freebies.
• Principal (E6 equivalent). The system design rounds shift toward roadmaps, multi-year evolution, and "how would you migrate the existing system to this design over 18 months." LPs are now a senior conversation about influence and judgment.

In this chapter the default target is SDE2 with notes on what tightens at SDE3.

5. What's different here

A few patterns specific to Amazon:

• They like operational thinking. Show monitoring, alarms, deployment, on-call. "How would you know this is broken in production?" is asked in almost every design round.
• They hate over-engineering. This is the most over-corrected thing candidates do. A senior engineer at Amazon ships a simpler thing on purpose. Designs that propose Kafka + microservices + multi-region failover for a small problem get marked down for poor judgment.
• They want strong opinions, weakly held. "I'd go with X. Here's why. If you can show me Y is better given Z, I'll flip — but right now, X." That sentence wins rounds.
• They take notes the whole time. Don't be thrown by it. The interviewer is writing because the feedback doc has to be detailed.
• Behavioral stories must be specific. "We had a bug, I fixed it" doesn't pass. Amazon wants the customer, the metric, the tradeoff, the alternative considered, and what happened next.

6. The Amazon-flavored mistakes

7. The five scenarios in this chapter

8. The next page

The catalog search. The system every senior engineer at Amazon has touched, the question they like to ask candidates because every answer reveals something about how the candidate thinks at scale.