RepostedAmazon logo

Amazon · Phone screen → Onsite

In process

Amazon Full-Time Interview Timelines: 9 Experiences (Part 2 of 2)

Software engineer · Full-time10 years experienceBachelor's degreeSep 2026
Shared by

Nine Amazon full-time loops from September 2026, most of them SDE II: AI repo debugging OAs, LeetCode-medium coding, LLD, system design and probing LPs.

At a glance

Coverage: 9 full-time reports from September 2026 (one phone screen on August 29): 5 explicitly SDE II / Software Engineer II, 4 with 3–10 years of experience and no level stated. United States, Seattle and Bangalore where stated.

In this series

  • Part 1: the overview, with question tables, patterns and common questions for all nine reports.
  • Part 2 (current): the nine candidate timelines.

What came up

Every candidate got Leadership Principles questions, often with three layers of follow-up on impact, users and AI use. Coding was mostly familiar LeetCode problems that grew with extensions.

9 detailed candidate timelines

01: offer: evolving BFS, return LLD, safest path and config sync

Report date: September 13, 2026 · Level: SDE II · Result: Offer

  1. OA: AI-assisted repository debugging and one DSA problem (not recalled).
  2. Round 1, DSA + LP: a custom BFS that gained constraints until it became multi-source BFS; the code was finished, with one known edge case left uncovered. LP: Learn and Be Curious, Ownership.
  3. Round 2, LLD + LP: design and partly implement a return drop-off reservation system (entities, interfaces, booking flow, store availability); follow-up: find nearby eligible stores, with a hint toward how ride-sharing finds nearby drivers. LP: Dive Deep, Invent and Simplify.
  4. Round 3, Bar Raiser: the safest path in a grid, optimized repeatedly. LP: Customer Obsession, Deliver Results.
  5. Round 4, hiring manager + HLD: multi-device configuration sync (APIs, data model, versioning, retries, idempotency, offline devices, conflicts), with failure scenarios added along the way. LP: Earn Trust; Have Backbone, Disagree and Commit.

02: an SDE II loop that ended in an SDE I offer

Report date: September 9, 2026 · Level: SDE II · Location: Bangalore · Experience: 3.3 years · Result: SDE I offer

  1. OA: a problem solved by binary search on the answer, and debugging a Python/Django project; the AI assistant would not reveal the bug, so most debugging was manual.
  2. Phone screen: Count Stepping Numbers in Range (solved with BFS, one undisclosed follow-up) and Capacity to Ship Packages Within D Days; LP: Bias for Action.
  3. Design (an HLD/LLD scheduling mix-up): Amazon inventory management, with a follow-up on two users ordering the last unit at once (race conditions, overselling, locks, transactions); LP: Earn Trust. Feedback: neutral.
  4. DSA: Container With Most Water, then Trapping Rain Water as a follow-up, and Cheapest Flights Within K Stops; LP: Bias for Action, Dive Deep. Feedback: very positive.
  5. Hiring manager + LLD: a distributed rate limiter (fixed window, sliding window, token bucket, Redis, locking, multiple instances, consistency); LP: Deliver Results. Feedback: positive.
  6. Bar Raiser: a knapsack-style problem; the greedy approach was challenged and the candidate moved to DP but did not finish; LP: Customer Obsession, Ownership. Feedback: negative.

03: two openings, a scheduling phone screen and a withdrawn onsite

Report date: September 2026 · Level: SDE II · Location: a Seattle team for the second opening · Result: Withdrew

  1. First opening, OA: a drone delivering packages clockwise around a loop with travel times between neighboring points (prefix sums), and AI-assisted debugging of a full project where reading the README's function definitions was enough. The role was filled before the phone screen, and Amazon offered three other openings.
  2. Second opening, phone screen (about 30 minutes BQ, 30 coding): a missed deadline, a self-started project and a high-impact project, each with follow-ups; then the vehicle-inspection registration scheduler: registration every six months, closed on holidays, one-hour slots from 10 a.m. to 4 p.m., spread n vehicles as evenly as possible; follow-up: change the window to 11 a.m.–3 p.m. The candidate advanced despite feeling it went poorly.
  3. Onsite, hiring manager: a deep dive on an AI-agent project (implementation, users, departments); biggest mistake, a process improvement, conflict and criticism; coding: Meeting Rooms, where the candidate first forgot to keep the maximum overlap during the sweep, then fixed it.
  4. Onsite, second interviewer: something new you learned and applied, your biggest benefit to the company, and a customer you made especially happy (no follow-ups); design a low-latency notification ingestion service for bursts, with clarifying questions answered "it is up to you".
  5. The candidate cancelled the next day's interviews and withdrew.

04: a palindrome phone screen

Interview date: August 29, 2026 (report date about September 6) · Level: Software Engineer II · Experience: four years · Result: Passed the phone screen

  1. Three behavioral questions, mostly Ownership and Dive Deep (wording not kept).
  2. A fairly simple palindrome coding problem (exact variant not kept), with approach, edge cases and complexity. The recruiter confirmed the pass the next day.

05: three technical prompts before the LPs

Report date: about September 18, 2026 · Level: "Amazon II" · Experience: three years · Result: Not stated

  1. LRU Cache (LeetCode 146), solved quickly, with two undisclosed follow-ups.
  2. Design a restaurant where many robots coordinate; a broad prompt with many follow-ups.
  3. Design a Kafka-based file watcher for millions of files.
  4. LP: a tight deadline where a project could not be finished in time; a Dive Deep root-cause story, told as debugging.

06: a Resilient Hub screen with a product manager and an SDE

Report date: September 2026 · Level: not stated · Result: Rejected

  1. BQ first: a time you took ownership, and a time you faced a challenge; follow-ups on AI use, impact, number of users and the manager's reaction.
  2. Coding: a latest-timestamp key-value object with set(timestamp, key, value) and get(key) returning the value with the greatest timestamp (no capacity or eviction). The candidate spent too long understanding it and did not finish.

07: three LPs left 15 minutes for Maximum Frequency Stack

Report date: about September 1, 2026 · Level: not stated · Experience: ten years · Result: Rejected the next day

  1. Three standard LP questions (wording not kept).
  2. With about 15 minutes left: Maximum Frequency Stack (LeetCode 895). The candidate saw the stack-of-stacks direction but gave an O(log n) approach and did not finish; the interviewer later pointed out a correctness issue.

08: a four-round onsite with follow-up-heavy coding and design

Report date: about September 20, 2026 · Level: not stated · Experience: four years · Applied: referral · Result: Pass per the post; the narrative said awaiting a result

  1. Four rounds with four interviewers: LeetCode-medium coding that kept gaining follow-ups, system design that drilled into why, scale, consistency and failure handling, and standard LPs tied to prepared project stories. No prompts were kept.

09: a first-round screen with conflict and tight-timeline questions

Report date: about September 3, 2026 · Level: not stated · Experience: four years · Applied: referral · Result: In process

  1. Only the first round was done: BQ on a conflict and a tight timeline; the technical prompt was not kept.

Keep practicing

In this series

Sources: Anywhere useful interview info can be found.

0 total reactions. Your reactions: none.
0 comments

Comments

0

No comments yet — be the first to share what you know.