DevOps Engineer Interview Experience at Amazon (Retail) – 2026

A recruiter call covering background and confirming the role sits within Amazon's retail and fulfillment infrastructure org specifically, distinct from AWS, with Leadership Principles assessed throughout every subsequent round rather than in one standalone interview.

Two coding problems of hard difficulty - one requiring an efficient approach to finding the shortest path in a weighted graph with an additional constraint on maximum number of edges used, expected to be solved with a modified Bellman-Ford tracking path length alongside distance, and one on interval scheduling to maximize the number of non-overlapping warehouse pickup slots, expected to be solved with a greedy approach sorting by end time.

Opens with a leadership-principle question about a time the candidate owned a project end to end despite it not being explicitly assigned to them, then moves into two coding problems - finding the intersection point of two singly linked lists using the two-pointer length-difference technique, and a knapsack-variant problem on maximizing value under a warehouse storage weight constraint, best solved with a bottom-up DP table after first explaining a brute-force recursive approach.

Candidates are asked to design an inventory-update system for a major sale event where write volume to product stock counts spikes dramatically and overselling a product is a real business risk. A strong approach is a hybrid model - strongly consistent stock decrements at the final checkout step to prevent overselling, but eventually-consistent stock-count displays elsewhere on the site to avoid every product-page view contending for the same strongly-consistent read path. Interviewers typically push on how to handle a brief window where displayed stock and true stock diverge during high load; a conservative buffer (displaying slightly lower stock than actual) during known high-load windows reduces the chance of a customer seeing 'in stock' for an item that sells out before checkout completes.

The most consequential round in the loop, per multiple candidates' accounts, combining a low-level design problem (designing a rate limiter class supporting multiple algorithms - token bucket and sliding window - behind a common interface) with an extensive leadership-principles discussion. The behavioral portion focuses heavily on a time the candidate disagreed with a technical decision made by someone senior to them and how they handled it - strong answers describe raising the disagreement with concrete data on why the decision would cause a specific downstream problem, and ultimately committing fully to the team's chosen direction once the decision is made despite the initial disagreement, tying directly to Amazon's 'Have Backbone; Disagree and Commit' principle.