Company interview guide

eBay Machine Learning Interview Guide (2026)

Practical Python and SQL coding, ML fundamentals and marketplace ML cases, with a heavy focus on your past projects.

eBay's machine learning interview loop has 8 reported stages, from recruiter or hr call to projects, hiring manager and behavioral; typical length: 2-10 weeks.

Roles covered: ML Engineer, Applied Scientist and Data Scientist. The question bank holds 4 questions reported from eBay interviews, 1 of them free to solve in your browser.

Try a free question

Interview process length
2-10 weeks
Stages
8
Reports linked to bank questions
4
Bank report dates
Feb 2017–Dec 2025
Questions in bank
4 (1 free)
Last reviewed
October 2026

How the eBay interview loop works

Usually a recruiter or HR call, then one or two technical rounds with the hiring manager or team members, then a final stage of about three to five interviews, run virtually or as a half-day panel. The technical rounds mix Python coding, ML questions and a deep dive on your projects; data scientist loops add SQL and sometimes statistics or product questions, and some teams send a take-home instead of a live round. Rounds reported with a length run 40 to 60 minutes. Candidate reports are short, several mention that the job posting did not match what was asked, and only a few describe a whole loop, so most stage details rest on a handful of accounts.

The eBay interview process, stage by stage

01Recruiter or HR call

A short phone call with a recruiter (about 20 min in one report); some teams start with the hiring manager instead

A short conversation about your background, why you want the role, location or relocation, and how soon you can start. A few candidates went straight to a hiring-manager call, and several say the team's real focus (for example NLP) only became clear later, so use this call to find out what the team works on.

What it tests

  • Why eBay and why this role
  • How your experience fits the team
  • Logistics: location, relocation, start date, languages and tools you use

How to prepare

  • Ask which team you would join and what it works on; job postings are often generic.
  • Ask for the number, length and type of each later round, since formats differ by team.
  • Have a two-minute summary of your most relevant ML project ready.

Recruiter, HR or phone screen as the first step in 9 ML, applied and data science reports since 2020 (5 since 2024); calls are short and about background and logistics.

02Coding round

40-60 min live in Python, in a screen or the final stage; some ML screens run on an online coding platform

Most coding questions are easy to medium: string and array manipulation, binary search, linked lists, recursion, with a talk about time and space complexity. One senior recommendations candidate faced a much harder binary-search problem in 40 minutes, and one applied-scientist round squeezed live coding into its last 15 minutes. Data scientist loops sometimes pair one Python problem with one SQL problem.

What it tests

  • Clean Python for array, string and search problems
  • Time and space complexity of your solution
  • Basic data structures such as linked lists
  • Working accurately under a short time limit

How to prepare

  • Drill easy and medium array, string and binary-search problems until they are quick.
  • State complexity unprompted; several reports say it was asked.
  • Expect coding to appear at the end of a mostly conversational round, and keep time for it.

Live coding in 12 ML, applied and data science reports since 2020 (7 since 2024), mostly easy to medium; one senior ML report had a much harder 40-minute problem.

03ML coding from scratch

About 60 min with a dataset: preprocess, train and evaluate a simple model in Python, often with project questions after

A hands-on round reported for both ML engineer and data scientist roles: clean a small dataset, fit a basic model and report its accuracy, or implement a classic algorithm such as k-means or logistic regression yourself. One candidate did not get the code running in time, so a simple working version beats an ambitious one.

What it tests

  • Data preparation: missing values, scaling, train and test split
  • Implementing k-means or logistic regression without a library
  • Choosing and computing evaluation metrics
  • Explaining what you would try next

How to prepare

  • Practise a full preprocess, train, evaluate loop in plain NumPy in under 30 minutes.
  • Get a baseline running first, then improve it.
  • Be ready to explain convergence and the loss of the algorithm you wrote.

Three reports since 2020 (two since 2024): a one-hour preprocess, train and evaluate task for an ML engineer, a k-means implementation and a logistic regression on a dataset.

04ML depth and breadth

A technical round or part of one, often led by the hiring manager or a team scientist

The most common technical content: how you would choose a model, tune it and evaluate it, classic topics such as the bias-variance trade-off and precision versus recall, and deep learning basics up to the transformer. Questions are often tied to your past models or to a problem the team has already solved, and some teams go deep on their own area, such as NLP.

What it tests

  • Model selection, hyperparameter tuning and evaluation
  • Classification metrics: precision, recall and their trade-off
  • Bias and variance, overfitting, regularisation
  • Deep learning basics: CNN versus RNN, transformer architecture
  • Depth in the team's own domain

How to prepare

  • Revise classical ML fundamentals as carefully as deep learning; reports ask both.
  • Be able to sketch a transformer block and explain its parts and cost.
  • Link each answer to a model you have actually built and how you measured it.

ML theory or breadth questions in 12 ML, applied and data science reports since 2020 (7 since 2024), from metrics and bias-variance to transformers; depth varies by team.

05SQL, statistics and product sense (data science)

Mainly data scientist loops: a SQL coding round or a combined SQL and product-sense round, sometimes with statistics questions

Data scientist candidates regularly get a SQL problem, sometimes alongside a Python one, and some rounds add product sense, statistics or experiment questions such as which KPIs to track and how to compare a test group with a control. One senior ML engineer was asked for a query in SQL or pandas. Older reports also include short probability puzzles.

What it tests

  • Writing correct SQL (joins, aggregation) under time pressure
  • Choosing KPIs for an e-commerce question
  • Comparing test and control groups with the right statistics
  • Probability and expected-value reasoning

How to prepare

  • Practise SQL and the same task in pandas; one interviewer accepted either.
  • Prepare a framework for marketplace metrics: buyers, sellers, listings, conversion.
  • Revise hypothesis tests and A/B test design basics.

Free reported questions for this stage

SQL, statistics or product questions in 8 reports since 2020, 7 of them data scientist roles (4 since 2024); probability puzzles appear in two 2017 reports.

06Take-home assignment (some teams)

Some teams only: a modelling task from 24 hours to 2 weeks, submitted as code plus a presentation or write-up

A minority of loops include a take-home, usually a classification problem on a provided dataset. Candidates handed in code with a presentation and sometimes a written report, and one ML engineer said presenting it clearly mattered. In one data science loop it replaced a planned technical interview.

What it tests

  • End-to-end modelling on a real dataset
  • Sound validation and metric choice
  • Clear presentation of results and trade-offs

How to prepare

  • Ask early whether your team uses a take-home and how long you get.
  • Spend real time on the presentation; it is part of what is judged.

Two reports since 2020 (a 24-hour classification task for data science, a two-week task for an ML engineer in 2024); two 2017 data science reports also had one.

07ML case and system design

Usually in the final stage, as an open discussion; some rounds are general distributed-system design

Case questions come from eBay's own business: how to find fraudulent listings, how to classify users, how to design a recommender system. Some ML engineer loops include system design that turns out to be general distributed design (sharding, latency, throughput) rather than ML design, so prepare for both.

What it tests

  • Framing a marketplace problem as an ML task with labels and metrics
  • Handling heavy class imbalance and costly errors
  • Recommender and ranking design
  • Scalability basics: sharding, latency, throughput

How to prepare

  • Prepare blueprints for listing fraud detection and for a marketplace recommender.
  • Always say how you would measure success offline and online.
  • Review general system design basics too; one ML loop asked only that.

ML case or system design in 7 ML, applied and data science reports since 2020 (5 since 2024); one ML engineer loop asked general distributed design instead.

08Projects, hiring manager and behavioral

Woven into most rounds, plus a hiring-manager or leadership conversation

Nearly every report includes detailed questions on your past projects, and the hiring manager often runs one round, sometimes as the first technical conversation. Final loops add a leadership or behavioral round with questions such as why eBay.

What it tests

  • Depth behind the projects on your resume
  • Your own contribution and the impact you measured
  • Motivation for eBay and the team
  • Collaboration and leadership

How to prepare

  • Prepare two projects end to end: problem, data, model, metric, result, what you would change.
  • If you joined a project midway, say so and focus on what you owned.
  • Have a specific answer to why eBay that names the team's problem space.

Project deep dives, hiring-manager or behavioral rounds in 14 ML, applied and data science reports since 2020 (8 since 2024); project questions are the most consistent element.

Questions and themes reported most often

  • Deep dive on a past project and the models you built (9 reports, projects, hiring manager and behavioral)
  • SQL coding problem in the data scientist loop (7 reports, sql, statistics and product sense (data science))
  • Easy-to-medium Python coding on strings, arrays or binary search (3 reports, coding round)
  • Implement a simple model from scratch: preprocess, train, evaluate (3 reports, ml coding from scratch)
  • Why eBay and why this role (3 reports, projects, hiring manager and behavioral)
  • Choosing, tuning and evaluating a model; precision and recall (2 reports, ml depth and breadth)
  • How would you identify fraudulent listings on eBay? (1 reports, ml case and system design)
  • Design a recommender system (1 reports, ml case and system design)

What eBay asks about

  • ML concepts & theory25%
  • General coding20%
  • ML case & system design15%
  • SQL, stats & product15%
  • Projects & behavioral15%
  • ML coding from scratch10%

eBay interview prep checklist

  • Two projects rehearsed end to end, with your own contribution and measured impact
  • Implemented k-means and logistic regression from scratch in NumPy
  • Solved a timed set of easy and medium array, string and binary-search problems
  • Can explain precision, recall, bias-variance and the transformer block clearly
  • Sketched designs for listing fraud detection and a marketplace recommender
  • Data science track: a set of SQL problems solved in SQL and in pandas
  • Asked the recruiter for the type and length of every round

Tips

  • Ask the recruiter which team you are interviewing for and what each round covers; reports often say the posting did not match the interview.
  • Prepare your projects in depth first; they come up in almost every round.
  • Keep coding fluency at the easy-to-medium level sharp, including complexity analysis.
  • Be able to write k-means or logistic regression and a preprocess, train, evaluate loop without libraries.
  • Revise classical ML fundamentals (metrics, bias-variance, tuning) as well as deep learning basics.
  • For data scientist roles, practise SQL alongside Python and basic A/B test statistics.
  • Think through marketplace ML problems: listing fraud, recommendations, search and buyer or seller segmentation.
  • Prepare for general system design questions as well as ML design if you are an ML engineer.

See the rest of the eBay loop

The later stages, the questions reported for them and how to prepare are in Pro.

Unlock with Pro

Practice the eBay loop

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

eBay interview FAQ

How many interview rounds does eBay have for ML roles?

Candidates report 8 stages: Recruiter or HR call, Coding round, ML coding from scratch, ML depth and breadth, SQL, statistics and product sense (data science), Take-home assignment (some teams), ML case and system design and Projects, hiring manager and behavioral. Usually a recruiter or HR call, then one or two technical rounds with the hiring manager or team members, then a final stage of about three to five interviews, run virtually or as a half-day panel. The technical rounds mix Python coding, ML questions and a deep dive on your projects; data scientist loops add SQL and sometimes statistics or product questions, and some teams send a take-home instead of a live round. Rounds reported with a length run 40 to 60 minutes. Candidate reports are short, several mention that the job posting did not match what was asked, and only a few describe a whole loop, so most stage details rest on a handful of accounts.

How long does the eBay ML interview process take?

Reported timelines run 2-10 weeks.

What does eBay ask in ML interviews?

By reported question mix: ML concepts & theory 25%; General coding 20%; ML case & system design 15%; SQL, stats & product 15%; Projects & behavioral 15%; ML coding from scratch 10%. The most repeated themes: Deep dive on a past project and the models you built; SQL coding problem in the data scientist loop; Easy-to-medium Python coding on strings, arrays or binary search.

How should I prepare for the eBay ML interview?

Ask the recruiter which team you are interviewing for and what each round covers; reports often say the posting did not match the interview. Prepare your projects in depth first; they come up in almost every round. Keep coding fluency at the easy-to-medium level sharp, including complexity analysis.

Are there free eBay interview questions?

Yes. LeetML has 4 questions reported from eBay interviews, and 1 of them are free to solve in the browser; they are listed by stage above.

Related interview guides

Official sources