Amazon SDE 1 interview questions by round

You want the SDE 1 list, not the generic Amazon list that reads like a senior loop. This section names the families that actually appear in the online assessment and the phone screen, which is where most new-grad candidates are filtered. Patterns transfer; a leaked prompt from last season usually does not.

• Arrays and strings. Two pointers, sliding windows, frequency counts, in-place rewrites.
• Hash maps. Grouping, anagrams, prefix structures, and deciding what to key on.
• Trees and graphs. BFS and DFS, including grids, connected components, and simple shortest paths.
• Stacks and queues. Monotonic stacks, sliding-window maximum, and order reconstruction.
• Sorting and heaps. K-th elements and merge shapes, not contest-hard geometry.

The assessment also includes a work-style survey mapped to Leadership Principles. Treat that survey as scored. Phone screens then ask you to write in a shared editor while talking, and they often close with one behavioral probe. Role-specific banks sit under the question banks topic.

Amazon will still phrase these as tell me about a time. At SDE 1 the stories are allowed to come from internships, class projects, campus jobs, and group work. The follow-ups are not softer. Interviewers want your action, the alternative you rejected, and a result you can defend.

• Tell me about a time you owned something that was not assigned to you.
• Describe a time you disagreed with a teammate. What did you do after the decision?
• Tell me about a time you missed a deadline or shipped something that broke.
• Give an example of a time you used data, not opinion, to change a plan.
• Describe a time you simplified a process or cut extra work without hurting quality.
• Tell me about the most complex bug or design choice in a project you owned.

Prepare 6 to 8 distinct stories, not one polished anecdote. Map each to two principles so you can rotate when two interviewers hunt the same theme. How the Loop uses those principles is covered on the company interviews topic.

Use STAR, then starve the first two letters. A short situation, one sentence of task, most of your time on action in the first person, and a result with a number if you have one. "We improved the pipeline" fails. "I cut the batch job from 40 minutes to 12 by partitioning the input" holds, even if it happened on an internship.

A new-grad candidate with one internship brought a caching story and got asked, in order, what she changed in the client, why not a longer TTL, what the hit rate was before, and what broke when keys collided. She had the numbers from her own dashboard, so the story held. Her second story was a group-course project in the plural, and the interviewer moved on in a minute.

Rehearse the follow-up half out loud. You can run that pressure against an AI interviewer on the mock interview page.

On a spoken Amazon Chime round, SubcueAI transcribes the interviewer's question and surfaces a suggested structure on your side, in the desktop overlay on macOS and Windows or in the Chromium browser extension side panel. No meeting bot joins the call, and nothing is injected into the meeting page. For Leadership Principles the useful output is which of your own stories fits, not a script.

Two honest limits. If the online assessment or a coding hour is proctored, recorded, or screen-shared, live help is out of scope. If you are asked to share your screen, anything on your screen is visible. More on how spoken help is architected sits in the detectability topic.