Amazon AI Engineer reports are limited. One candidate failed a stack-based online assessment, and another described a phone screen followed by a five-round virtual loop of coding, ML and leadership questions.
$163K
Avg. Base Comp
$220K
Avg. Total Comp
5 rounds
Typical Rounds
Not reported
Process Length
Only two reports for this exact role support this guide, and they describe different stages, so read them as separate snapshots instead of one fixed process.
One candidate was rejected at the initial online assessment. The first problem was a stack-based coding question in the style of Valid Parentheses, where the challenge was recognizing the right data structure and handling edge cases. They lost time on an overly complicated first approach and only felt comfortable after breaking the problem into smaller cases. Afterward they practiced Min Stack and Daily Temperatures, and concluded that fast pattern recognition matters as much as understanding each concept.
The second candidate, who did not receive an offer, described a phone screen followed by a five-round virtual loop that stretched over about a month. Each round ran roughly an hour: two coding rounds, an ML application round, an ML breadth round and an ML depth round. Two leadership principle questions were embedded in every round. The coding was LeetCode-style and leaned on CS fundamentals. The ML application round was built around a real use case for the target team. The breadth and depth rounds covered improving CLIP, LLM benchmarks, GRPO, RoPE embeddings and general LLM architecture.
This candidate also found that only one interviewer came from the target team, and that NLP and statistics questions went beyond what the job description and Amazon's preparation materials highlighted. Prepare broadly, and keep concise leadership stories ready for every round.
Synthesized from 5 candidate reports by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Amazon process.
Share your own interview experience to unlock all reports, or subscribe for full access.
Sourced from candidate reports and verified by our team.
Topics based on recent interview experiences.
Featured question at Amazon
Write a query that returns all neighborhoods that have 0 users.
| Question | |
|---|---|
| Paired Products | |
| 2nd Highest Salary | |
| Upsell Transactions | |
| Longest Streak Users | |
| Rolling Bank Transactions | |
| Customer Orders | |
| Comments Histogram | |
| Top Three Salaries | |
| Closest SAT Scores | |
| Get Top N Frequent Words | |
| Subscription Overlap | |
| Monthly Customer Report | |
| Size of Joins | |
| Random SQL Sample | |
| Merge Sorted Lists | |
| Manager Team Sizes | |
| Download Facts | |
| Top 3 Users | |
| Average Quantity | |
| Month Over Month | |
| 7 Day Streak | |
| Cumulative Reset | |
| Compute Deviation | |
| Always Excited Users | |
| Third Purchase | |
| Duplicate Rows | |
| Max Quantity | |
| Flight Records | |
| Cumulative Sales Since Last Restocking |
Synthesized from candidate reports. Individual experiences may vary.
One candidate's initial online assessment opened with a stack problem similar to Valid Parentheses. It required spotting the right data structure and reasoning through edge cases. They lost time on a complicated first approach, then improved by splitting the problem into smaller cases, but did not finish confidently and was rejected at this stage.
A different candidate, who reached the full loop, reports a phone screening before the virtual rounds. The report does not describe what the screen covered, so there is no evidence for its format or questions. It is the step that came before the five-round loop in that account.
The loop candidate had two coding rounds of roughly an hour each. The problems were LeetCode-style algorithmic questions that tested CS fundamentals more than ML-specific optimization. After failing the assessment, the other candidate practiced stack patterns such as Min Stack and monotonic-stack problems and stressed recognizing patterns quickly.
The loop included an ML application round grounded in a real use case for the target team, plus breadth and depth rounds tailored to the candidate's background. Questions covered research directions for improving CLIP, LLM benchmarks the candidate had used, GRPO, RoPE embeddings and general LLM architecture. NLP and statistics topics also came up.
The loop candidate reports two leadership principle questions in every one of the five rounds. They found it repetitive but consistent with Amazon's culture focus. Because the questions recur in each round, prepare several distinct stories rather than relying on one or two examples.
