Amazon Data Scientist interview typically runs 2-5 rounds: HR screen, technical rounds, and final loop/bar raiser. Timeline is about 2-6 weeks, with broad Amazon Leadership Principles woven through the process.
$135K
Avg. Base Comp
$255K
Avg. Total Comp
4-6
Typical Rounds
2-6 weeks
Process Length
We've seen Amazon's Data Scientist loop behave less like a single statistics exam and more like a pressure test for candidates who can move between analytics, ML, experimentation, and ownership. Candidate reports cluster around a few repeated signals: timed SQL or coding, ML and statistics fundamentals, and case prompts where the clean experiment is unavailable. The most useful preparation target is structured breadth under ambiguity: you need to name assumptions, choose a method, and explain what would make your conclusion unreliable.
Amazon also makes the behavioral bar part of the technical evaluation. Leadership Principles show up inside SQL, ML, and causal-inference conversations, especially through project deep dives and final-loop cases. A strong answer usually connects the technical choice to customer or business impact, then defends tradeoffs without hiding behind jargon. Reports mentioning out-of-stock behavior, causal inference without A/B tests, and team-specific case discussions all point to the same thing: Amazon is testing owner-level judgment, not just model vocabulary.
For prep, don't treat this as a checklist of isolated topics. Practice switching from SQL to experimentation to model evaluation in one sitting, and pair each technical answer with a short business rationale. The candidates who sound most credible are the ones who can say what they'd measure, what could bias the result, and how they'd communicate the decision if the data is messy.
Synthetized from 13 candidates 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.
The HR call set expectations for a phone screen followed by four to six virtual onsite rounds. The phone screen was supposed to include two data manipulation questions, machine learning questions, and a leadership-principles behavioral question. The interviewer did not end up asking the behavioral question, but I felt confident about my machine learning and SQL answers.
In the phone screen, I was asked general machine learning questions such as what PCA is, how to choose dimensions, what L1 regularization is, and the difference between random forests and XGBoost tree models. The SQL question was a moving average problem.
The virtual onsite had four rounds. The first round focused on project details, including features, why I chose a particular model, and the workflow. The behavioral questions covered pushing back on a customer request under customer obsession and describing an innovative project under invent and simplify.
The second round was coding. It included data wrangling in Python with list manipulation and pandas operations such as groupby, rolling average, pivot, and merge. The behavioral question was about disagreeing and committing.
The third round was a hiring-manager case study tied to the team's project area. I needed to clarify the problem and simplify my approach. The behavioral questions covered digging deep and handling an unexpected obstacle.
The fourth round tested machine learning breadth. Topics included bias-variance tradeoff, overfitting, linear regression assumptions, normal distributions, missing data, boosting versus bagging, and model metrics. The behavioral questions covered improving an already good system and taking on work outside my formal responsibility.
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 | |
|---|---|
| 2nd Highest Salary | |
| Rolling Bank Transactions | |
| Customer Orders | |
| Comments Histogram | |
| Closest SAT Scores | |
| Subscription Overlap | |
| Top Three Salaries | |
| Upsell Transactions | |
| Monthly Customer Report | |
| Merge Sorted Lists | |
| Compute Deviation | |
| Experiment Validity | |
| Download Facts | |
| Average Quantity | |
| Random SQL Sample | |
| Manager Team Sizes | |
| Button AB Test | |
| Month Over Month | |
| Flight Records | |
| Prime to N | |
| Paired Products | |
| Swipe Precision | |
| Top 3 Users | |
| Longest Streak Users | |
| Recurring Character | |
| Bank Fraud Model | |
| Jars and Coins | |
| Always Excited Users | |
| Project Pairs |
Synthesized from candidate reports. Individual experiences may vary.
The process typically starts with an HR or recruiter call to review your background, role fit, and Amazon Leadership Principles. This stage can be a relaxed alignment conversation, but candidates should still be ready to walk through past projects and explain why the role fits their experience.
Many candidates receive an online assessment before live interviews. It commonly includes SQL or coding questions, sometimes LeetCode-style problems, plus a leadership principles section and basic data science stack skills such as numpy, pandas, or practical data manipulation.
The next step is often a phone screen with a hiring manager or data scientist. This round usually mixes resume deep dive, behavioral questions, and technical fundamentals such as SQL, Python coding, ML basics, statistics, or discussion of past projects.
Some candidates are given a take-home style project and later present their solution. The discussion focuses on the reasoning behind the approach, the difficulties encountered, and how clearly the candidate explains tradeoffs, assumptions, and results.
The main interview loop is usually a series of back-to-back rounds covering coding, SQL, machine learning breadth and depth, statistics, behavioral questions, and project discussion. Candidates described a broad mix of topics including deep learning, causal inference, system design, and practical evaluation or experimentation cases.
The loop often includes a bar raiser, hiring manager case, or team-specific final discussion. This stage is heavily focused on Leadership Principles and role fit, while still probing how the candidate would approach realistic business, causal inference, or impact-estimation problems end to end.