Company interview guide

Etsy Machine Learning Interview Guide (2026)

Marketplace-flavored data science loops built on A/B testing, product metrics and live SQL, plus coding and ML basics for ML engineers.

Etsy's machine learning interview loop has 7 reported stages, from recruiter or hr screen to final panel: competencies, behavioral and manager; typical length: About 3-4 weeks when it runs smoothly, often longer.

Roles covered: Data Scientist and ML Engineer. The question bank holds 2 questions reported from Etsy interviews.

Try a free question from this loop's topics

Interview process length
About 3-4 weeks when it runs smoothly, often longer
Stages
7
Reports linked to bank questions
3
Bank report dates
Feb 2022–Jun 2026
Questions in bank
2 (0 free)
Last reviewed
October 2026

How the Etsy interview loop works

A recruiter or HR screen, then a call with the hiring manager, then one technical round, then a final panel of about four interviews of 30 to 60 minutes. For data scientists the technical round is live SQL, sometimes paired with interpreting a dashboard, and both the manager call and the panel lean on A/B testing and product case questions. For ML engineers the technical round is a one-hour coding interview, and the final stage mixed coding, ML fundamentals, ML design and a manager round; some older research-heavy loops ended with a talk on past work. This outline rests on about 20 candidate reports, only four of them from 2024 onward, and several candidates describe slow scheduling and late or missing updates, so plan for delays and ask the recruiter what each round covers.

The Etsy interview process, stage by stage

01Recruiter or HR screen

Short phone or video call with a recruiter

A non-technical call about your background, the role and the shape of the process. One candidate says the recruiter also walked them through the final panel ahead of time. Several reports describe slow scheduling and follow-ups that never came, so confirm next steps and timing before you hang up.

What it tests

  • A clear summary of your background and why the role fits
  • Interest in Etsy and its two-sided marketplace
  • Logistics: timeline, location, level

How to prepare

  • Ask what each later round covers; one candidate was told to expect statistics and programming and got a SQL questionnaire instead.
  • Agree on a date for the next update and follow up politely if it passes.

Recruiter or HR screen as the first step in 7 data science and ML engineer reports since 2019 (one since 2024); calls are about background and the role.

02Hiring manager call

About 45-60 min with the hiring manager, usually by video

A conversation about your experience and the team, but often more than that: data science managers have asked A/B testing, product scenario and SQL questions here, and one ML engineer was told the call was not technical and then got a live medium-level coding problem plus ML basics. Treat it as a technical round unless told otherwise.

What it tests

  • Depth behind the projects on your resume
  • Experiment design and product recommendations from a scenario
  • Basic ML concepts and your tech stack
  • Sometimes a live coding or SQL problem

How to prepare

  • Prepare a crisp walkthrough of one or two projects with the metrics they moved.
  • Have an editor or SQL environment ready in case the call turns technical.

Hiring-manager round early in 8 data science and ML engineer reports since 2017 (two since 2024); in three it included case, SQL or coding questions.

03Data science technical round: SQL and dashboard

Data scientist track: live SQL in a shared editor or online test tool, sometimes after interpreting a dashboard; can also appear in the final panel

The most consistent technical element for data scientists is live SQL at a basic to moderate level: joins, grouping and aggregation. Two 2023 reports also describe being shown a dashboard and asked to draw insights and propose follow-up questions. One 2025 candidate got the answers right but was pushed on writing the grouping in the interviewer's preferred way, so explain your query choices.

What it tests

  • Join types and when each changes the result
  • GROUP BY, aggregation and filtering on grouped results
  • Reading a business dashboard and turning it into questions
  • Explaining a query clearly as you write it

How to prepare

  • Drill short SQL problems until joins, GROUP BY and HAVING are automatic; questions are described as simple.
  • When given a dashboard, state the trend, a likely cause and the next cut of the data you would check.

Live SQL in 7 data scientist reports since 2019 (three since 2024), described as simple; a dashboard interpretation round appears in two 2023 reports.

04Coding screen and coding rounds

ML engineer track: about 60 min of live coding, at the phone screen and again in the onsite; you choose the language

ML engineer and research-leaning candidates describe medium-level algorithm problems, such as finding every run of consecutive integers that sums to k, decoding a nested string or converting Roman numerals, plus implementing an ML algorithm like k-means in plain Python. Interviewers were described as giving hints when stuck. These reports date from 2018 to 2022, so confirm the current format.

What it tests

  • Medium-level array and string problems with an optimal solution
  • Implementing a classic ML algorithm without libraries
  • Thinking out loud and handling edge cases
  • Passing the interviewer's test cases

How to prepare

  • Practise two-pointer and sliding-window problems; the reported k-sum problem has an O(n) solution.
  • Be able to write k-means from scratch in plain Python, including initialization and a stopping rule.

Live coding in 6 ML engineer and research-leaning reports from 2018 to 2022, usually one hour at the screen and one onsite round; none since 2023.

05A/B testing and product case

Data scientist track: case questions in the manager call and across the final panel, 30-45 min each

The most reported topic for data scientists. Expect to design an experiment for a feature, pick success and guardrail metrics for a marketplace with buyers and sellers, and make a product recommendation from a scenario. One candidate said most panel interviewers asked nearly the same questions, so a solid framework pays off more than once.

What it tests

  • Designing an A/B test: unit, metrics, duration and sample size
  • Choosing metrics for Etsy or for a specific product feature
  • Making a recommendation from ambiguous scenario data
  • Marketplace effects that complicate experiments

How to prepare

  • Have a repeatable experiment-design outline and practise saying it in two minutes.
  • Think about both sides of the marketplace when choosing metrics: buyers, sellers and Etsy itself.

A/B testing, metrics or product case questions in 7 data scientist reports since 2020 (three since 2024); the most consistent topic on this track.

06ML fundamentals and ML design

ML engineer and research-leaning tracks: ML basics at a phone screen or onsite round, plus open-ended ML design in some onsites

Questions are described as basic but interesting: the objective of a linear SVM, L1 versus L2 regularization and their geometry, k-nearest neighbours, and supervised versus unsupervised learning. One 2019 onsite had two ML design rounds, such as designing a recommendation system, and a 2022 intern loop included system design.

What it tests

  • Loss functions and objectives of classic models (SVM, regularized regression)
  • L1 versus L2 regularization and why L1 gives sparsity
  • Nearest-neighbour methods and clustering
  • Designing a recommender end to end

How to prepare

  • Review classic ML at whiteboard depth: be ready to write an objective and sketch the geometry.
  • For design, prepare a recommender outline with candidate retrieval, ranking and offline and online metrics.

ML basics in 7 ML engineer and research-leaning reports from 2017 to 2022; ML design in only two of them (a recommender, an intern system design).

07Final panel: competencies, behavioral and manager

About four interviews of 30-60 min in one day, virtual or onsite; older research-leaning loops added a talk on past work

The last stage mixes case or technical questions with core-competency and behavioral questions about how you work with stakeholders. For data scientists it is case-heavy, and one candidate said it was mostly behavioral; for ML engineers it included coding, ML and a manager round. Three reports from 2017 to 2019 also included a presentation of the candidate's research.

What it tests

  • Working style and stakeholder communication
  • Consistency of your answers across several interviewers
  • Depth behind past projects and how you would improve them
  • For research-leaning roles, presenting and defending your work

How to prepare

  • Prepare four or five stories on collaboration, influence and handling disagreement with stakeholders.
  • If a talk is requested, keep it to a clear problem, method, result and what you would change.

A final panel of about four interviews in 8 data science and ML engineer reports since 2017 (one since 2024); a research talk appears in three of them.

Questions and themes reported most often

  • Live SQL on joins and aggregation (7 reports, data science technical round: sql and dashboard)
  • Classic ML basics: SVM objective, L1 vs L2, kNN, supervised vs unsupervised (5 reports, ml fundamentals and ml design)
  • Deep dive on your resume and past projects (5 reports, hiring manager call)
  • Design an A/B test for a product change and read out the result (4 reports, a/b testing and product case)
  • Talk on your research after the onsite rounds (3 reports, final panel: competencies, behavioral and manager)
  • Choose success metrics for Etsy or a marketplace feature (2 reports, a/b testing and product case)
  • Interpret a dashboard and propose follow-up questions (2 reports, data science technical round: sql and dashboard)
  • Find every run of consecutive integers that sums to k (1 reports, coding screen and coding rounds)

What Etsy asks about

  • A/B testing & product metrics30%
  • SQL & data analysis20%
  • ML fundamentals17%
  • Coding15%
  • Behavioral & project deep dive13%
  • ML system design5%

Etsy interview prep checklist

  • Can design an A/B test end to end, including metrics, duration and marketplace pitfalls, in a few minutes
  • Have a metric framework for a two-sided marketplace feature
  • Solved a set of join and GROUP BY problems without hesitation
  • Practised reading a dashboard aloud and proposing next questions
  • For ML roles: solved the consecutive-run sum problem in O(n) and wrote k-means from scratch
  • Reviewed SVM, regularization and nearest-neighbour basics
  • Prepared project deep dives and behavioral stories about working with stakeholders

Tips

  • For data science roles, make A/B testing and metric selection your first priority; it comes up in the manager call and again across the panel.
  • Practise metric questions from both sides of a marketplace: buyer conversion, seller success and long-term health.
  • Keep basic SQL sharp and explain why you group or join the way you do, not just that the result is right.
  • For ML engineer roles, expect a medium-level coding problem even in a call described as non-technical.
  • Review classic ML at whiteboard depth: SVM objective, L1 versus L2, kNN and k-means.
  • Expect similar questions from several panel interviewers and answer each one fully rather than referring back.
  • Ask the recruiter for the content of every round, since some candidates were told one thing and asked another.
  • Budget extra time for scheduling and follow up if updates stall; several candidates had to chase for news.

Practice the Etsy loop

Work the stages in order: every reported question for this company, grouped by round, with solutions in the app.

Etsy interview FAQ

How many interview rounds does Etsy have for ML roles?

Candidates report 7 stages: Recruiter or HR screen, Hiring manager call, Data science technical round: SQL and dashboard, Coding screen and coding rounds, A/B testing and product case, ML fundamentals and ML design and Final panel: competencies, behavioral and manager. A recruiter or HR screen, then a call with the hiring manager, then one technical round, then a final panel of about four interviews of 30 to 60 minutes. For data scientists the technical round is live SQL, sometimes paired with interpreting a dashboard, and both the manager call and the panel lean on A/B testing and product case questions. For ML engineers the technical round is a one-hour coding interview, and the final stage mixed coding, ML fundamentals, ML design and a manager round; some older research-heavy loops ended with a talk on past work. This outline rests on about 20 candidate reports, only four of them from 2024 onward, and several candidates describe slow scheduling and late or missing updates, so plan for delays and ask the recruiter what each round covers.

How long does the Etsy ML interview process take?

Reported timelines run About 3-4 weeks when it runs smoothly, often longer.

What does Etsy ask in ML interviews?

By reported question mix: A/B testing & product metrics 30%; SQL & data analysis 20%; ML fundamentals 17%; Coding 15%; Behavioral & project deep dive 13%; ML system design 5%. The most repeated themes: Live SQL on joins and aggregation; Classic ML basics: SVM objective, L1 vs L2, kNN, supervised vs unsupervised; Deep dive on your resume and past projects.

How should I prepare for the Etsy ML interview?

For data science roles, make A/B testing and metric selection your first priority; it comes up in the manager call and again across the panel. Practise metric questions from both sides of a marketplace: buyer conversion, seller success and long-term health. Keep basic SQL sharp and explain why you group or join the way you do, not just that the result is right.

Are there free Etsy interview questions?

LeetML has 2 questions reported from Etsy interviews. The stage sections above link the free practice problems for the topics each round covers; the reported-question set itself is in the app.

Related interview guides

Official sources