Amazon Data Scientist Interview Guide

The Amazon data scientist interview leans on cultural fit more than almost any other FAANG data science interview, and Amazon defines that fit through its Leadership Principles. Behavioral evaluation reaches into every round, with each interviewer assigned specific Principles to assess you against, whether the conversation is about a SQL query, a model choice, or a past project. At the senior level, ownership and quantified impact often decide the final level.

This guide breaks down each stage of the Amazon data scientist interview, what interviewers look for, and how to prepare with example questions, actionable tips, and resources.

The Amazon data scientist interview process

Amazon's data science interview moves from a recruiter screen through a technical screen to a five-round onsite, with the technical bar centered on SQL, machine learning fundamentals, and project depth.

Here's what the interview process can look like:

• Recruiter screen: A 30-minute call covering work history, level fit, and motivation
• Technical screen: A 45-minute SQL and statistics interview over Amazon Chime
• Onsite loop: Five one-hour rounds covering data manipulation and scripting, data science breadth, data science depth, a hiring manager behavioral, and a Bar Raiser

Recruiter screen

The Amazon data scientist recruiter screen is a standard 30-minute call that confirms your background, level fit, and motivation. Your interviewer may be a recruiter or the hiring manager, and some teams run a second screen at this stage.

Amazon recruiters ask for your compensation expectations more directly than most, often early in the call. Unless you're holding competitive offers, avoid naming a number first.

Interviewers look for:

• Level fit: Whether your experience matches the seniority of the role
• Motivation: Why you want to work at Amazon and on this team
• Relevant background: How your work history maps to the role's core skills
• Communication: How clearly and concisely you summarize your background

Recently asked questions

Here are some real interview questions and prompts reported by candidates:

• Tell me about your resume.
• What's your experience with SQL?
• Why Amazon?

Technical screen

Amazon's data scientist technical screen is a 45-minute interview, generally over Amazon Chime, focused on SQL and statistics. You'll solve SQL challenges in a shared notepad against a question bank shared across data science teams.

Some teams, generally AWS government operations, add a take-home to this round. If a take-home is part of your loop, you'll pick one of three machine learning challenges across classification, NLP, or computer vision, and submit a solution.

Interviewers look for:

• SQL correctness: Whether your queries return the right result, ahead of optimization
• Window functions: Command of ranking, lag, and partitioning
• Statistical reasoning: How you work through probability and statistics prompts
• ML evaluation fluency: Familiarity with metrics like precision and AUC-ROC
• Pipeline thinking: How you'd build and analyze large-scale data pipelines

Recently asked questions

Here are some real interview questions and prompts reported by candidates:

• Given a retail dataset, find the best-selling item for each month, calculated as (unit price × quantity), and output the month, item description, and amount paid.
• Write a SQL query to explain month-to-month user retention from a set of user data.
• Name some common window functions.
• Estimate disease probability in one city when the nationwide probability is very low: if 1,000 people all test negative, what's the probability of disease in that city?
• There are 4 red balls and 2 blue balls; what's the probability of the two picks not matching?
• How do you inspect missing data, and when does it matter?

Data manipulation and scripting round

The Amazon data scientist data manipulation and scripting round opens the onsite with three SQL challenges that escalate in difficulty. Depending on your team, you may also get a second set of domain-specific challenges, such as graph traversal and route optimization for a logistics team.

Data-heavy teams, such as video or AWS, may add Python, PySpark, and PyTorch. Expect abstract questions on imbalanced-data scenarios and techniques like SMOTE, undersampling, and cost-sensitive learning.

Interviewers look for:

• SQL depth: How you handle escalating aggregation, self-joins, and multi-join CTE challenges
• Domain modeling: How you adapt to team-specific challenges like route optimization
• Imbalanced-data handling: Your command of techniques for skewed, real-world datasets
• Explainability: How you justify and interpret your model decisions
• Scripting fluency: Your comfort with Python and PySpark on large datasets, for data-heavy teams

Recently asked questions

Here are some real interview questions and prompts reported by candidates:

• How do you use PySpark on a large dataset?
• Given a table of (id, category, value) where each id has up to three categories, find the ids where two or more category values match.
• Given a 2 GB .csv with id and quantity columns and 50 million records, write a program to aggregate the quantity column.
• Design an A/B test for a green vs. yellow buy button.
• Design an experiment to evaluate Prime Video recommendations for both user satisfaction and business KPIs.
• How do you handle categorical variables?
• How do you interpret OLS regression results?

Data science breadth round

The Amazon data scientist breadth round tests your range across data science topics, usually splitting into a set of machine learning questions and a case study.

The case study scenario is often a generic prompt unrelated to Amazon's business, so treat it as a test of your end-to-end method more than your familiarity with Amazon's business. You'll walk the interviewer through your process: feature generation, model selection and reasoning, metric definition, and next steps.

Technical follow-ups and Leadership Principle questions also run throughout this round. LLM and GenAI concepts occasionally surface inside the machine learning questions.

Interviewers look for:

• Breadth of knowledge: Command across regularization, clustering, and boosting methods
• End-to-end reasoning: How you move from features to model selection to metrics
• Model justification: Why you select a given model for the scenario
• Metric definition: How you define and defend success metrics
• Leadership Principle signal: How your examples map to Amazon's Principles

Recently asked questions

Here are some real interview questions and prompts reported by candidates:

• Predict housing prices end-to-end, including feature generation, model selection, and metric definition.
• What's the difference between L1 and L2 regularization?
• What's bootstrapping?
• What's the difference between bagging and boosting?
• How would changing Prime membership fees affect the market?
• We have two models, one at 85% accuracy and one at 82%; which do you pick and why?
• How would you improve a classification model with low precision?

Data science depth round

The Amazon data scientist depth round is a project deep-dive that examines a past project end-to-end and the measurable business impact you drove. This round is where senior candidates most often get downleveled.

Lead with impact, ideally revenue or a clear org-level metric. At the senior level, interviewers look for evidence you scoped an ambiguous challenge yourself and drove the solution.

Interviewers look for:

• Measurable impact: The quantified business outcome of your project, ideally revenue
• Ownership and scope: Evidence you led an ambiguous project end-to-end
• Senior-level signal: Proof you operated as the senior data scientist, including mentorship
• Forward thinking: The mechanisms you built to prevent recurrence and guide future work
• Relevance and focus: Whether you stay on the decisions that mattered, past data-cleanup detail

Recently asked questions

Here are some real interview questions and prompts reported by candidates:

• Tell me about a recent project you executed end-to-end.
• What areas of ambiguity did you encounter, and how did you address them?
• How did you communicate goals and updates to non-technical stakeholders?

Hiring manager round

The Amazon data scientist hiring manager round asks behavioral questions but expects the same project detail as the depth round. Ground each answer in a specific past project, and connect your decisions to Amazon's Leadership Principles where they fit.

Because interviewers compare notes across the loop, draw each answer from a different project. Repeating one project across rounds signals limited range and can lead to a downlevel.

Interviewers look for:

• Behavioral depth: Project detail behind each behavioral answer
• Leadership Principle alignment: How naturally your examples map to Amazon's Principles
• Range: Distinct projects across rounds, with no single story reused
• Communication: How clearly you present decisions and results to a manager

Recently asked questions

Here are some real interview questions and prompts reported by candidates:

• Tell me about a project where you simplified something abstract for stakeholders.
• Tell me about a time a project deadline moved earlier than expected. How did you handle it, and what was the result?
• How would you improve a past project if you had more time?
• Tell me about a time you used data to build a recommendation and how you presented your findings.

Bar Raiser round

The Amazon data scientist Bar Raiser round is often the toughest in the loop, led by a trained interviewer from outside your team who focuses on Leadership Principles. The Bar Raiser is the one person in your loop with no connection to the team you're joining.

Bar Raisers work quickly and will interrupt if your answer loses focus. Like every Amazon interviewer, they're assigned one to three Principles to assess.

Interviewers look for:

• Leadership Principle evidence: Specific stories that map to the principles they're assigned
• Judgment under ambiguity: How you make sound decisions when the situation is unclear or incomplete
• Depth under follow-up: Whether your examples hold up under follow-ups
• Authenticity: Genuine, self-aware reflection in your answers
• Long-term potential: Signal that you raise the hiring bar

Recently asked questions

Here are some real interview questions and prompts reported by candidates:

• Tell me about a project that wasn't successful. What would you do differently?
• Tell me about a time you applied judgment to a decision when data wasn't available.
• Describe a time you kept an approach simple or saved on expenses.
• Describe a disagreement with your manager and how you handled it.
• Tell me about your most significant accomplishment.

How to prepare for the Amazon data scientist interview

1. Build 16 Leadership Principle stories: Amazon assesses the Principles in every round, so prepare multiple distinct examples with clear scope, ownership, and quantified impact.
2. Lead every story with impact: Open with business outcomes, ideally revenue or an org-level metric, then explain your approach and tradeoffs.
3. Vary your projects: Interviewers share notes, so line up at least three distinct projects and avoid repeating one across rounds.
4. Sharpen SQL skills: Prioritize correctness on aggregations, CTEs, subqueries, and window functions like ranking and lag before you optimize anything.
5. Prepare for explainability: Be ready to justify model choices with frameworks like SHAP and LIME.
6. Study recently asked questions: Work through recently asked Amazon data scientist questions to calibrate difficulty.
7. Don't rely on AI tools during interviews: Amazon prohibits candidates from using GenAI assistance during live interviews and assessments, and interviewers are trained to watch for it.
8. Practice with mock interviews: Run realistic, timed mock interviews to sharpen your SQL and Leadership Principle answers under time pressure. For targeted feedback, work through a loop with an expert coach.

About the Amazon data scientist role

Amazon data scientists connect the company's technical systems to customer-facing decisions, building models and pipelines that inform leadership across the business. Much of the work draws on internal systems to create analyses, build models, and drive business intelligence engineering (BIE) observations across business and technical domains.

Senior Amazon data scientists typically work on:

• Machine learning solutions for forecasting
• A/B testing and experimentation
• SQL queries and data pipeline development
• Analytical models and business intelligence solutions
• Mentoring junior data scientists

Amazon data scientist experience requirements

Senior Amazon data science roles expect end-to-end ownership: scoping ambiguous challenges, choosing methodology and tools, and leading solutions independently. A lack of mentorship experience is one of the most common reasons candidates get downleveled in this loop.

Additional resources

• Data Science interview course
• SQL interview course
• ML concepts for data scientists
• Python coding for data scientists
• Amazon Leadership Principles guide
• Amazon interview questions
• Amazon interview experiences

FAQs about the Amazon data scientist interview

How should you prepare for the Amazon data scientist interview?

Prepare for the Amazon data scientist interview by treating Leadership Principle stories as seriously as your technical prep. Build multiple distinct project examples with clear scope, ownership, and quantified impact, and practice SQL fundamentals like aggregations, CTEs, and window functions. Review machine learning breadth topics and be ready to justify model choices with explainability frameworks like SHAP and LIME.

How many rounds are in the Amazon data scientist interview?

The Amazon data scientist interview typically runs seven rounds: a recruiter screen, a technical screen, and a five-round onsite. The onsite covers data manipulation and scripting, data science breadth, data science depth, a hiring manager behavioral, and a Bar Raiser. Most loops take 4-6 weeks, and senior loops often run closer to 8 weeks.

How important are the Leadership Principles in the Amazon data scientist interview?

Leadership Principles are central to the Amazon data scientist interview and surface in every round, including the technical ones. Each interviewer is assigned one to three Principles to assess, and the Bar Raiser round is built largely around them. Strong technical candidates who can't map their experience to the Principles are routinely downleveled or rejected.

If you're rejected, how long before you can reapply to Amazon?

Amazon's cooling-off period depends on how your loop ended. A near-miss, which recruiters reportedly flag internally as a "recycle," often carries a wait of around six months, while a clearer rejection can mean a year or more before you reapply to a similar role. The freeze usually applies to the specific team or role, so you can often interview with a different Amazon org sooner, or with more than one team at once.

How much does an Amazon data scientist make?

Here are the reported compensation ranges by level for Amazon data scientists, according to Levels.fyi:

• L4 (Data Scientist I): ~$189K
• L5 (Data Scientist II): ~$282K
• L6 (Data Scientist III): ~$382K
• L7 (Principal Data Scientist): ~$625K

These figures combine base salary, equity, and bonus. Amazon back-loads equity on a 5/15/40/40 schedule, concentrating about 80% of it in years three and four.