Company interview guide
Shopify Machine Learning Interview Guide (2026)
Practical pair programming with AI allowed, a signature Life Story conversation, and commerce-flavoured ML design and modeling.
Shopify's machine learning interview loop has 9 reported stages, from recruiter or hiring-manager call to team placement & offer; typical length: About 1-3 months.
Roles covered: ML Engineer, Data Scientist and AI Engineer. The question bank holds 15 questions reported from Shopify interviews, 1 of them free to solve in your browser.
- Interview process length
- About 1-3 months
- Stages
- 9
- Reports linked to bank questions
- 15
- Bank report dates
- Aug 2025–Sep 2026
- Questions in bank
- 15 (1 free)
- Last reviewed
- October 2026
How the Shopify interview loop works
A short recruiter or hiring-manager call (engineering-track candidates sometimes skip it), an online assessment, then a 75-minute pair-programming round that gates the final loop. The final loop is usually four interviews: Life Story, a project deep dive, ML system design and ML modeling for ML engineers, or SQL, analysis and product-case rounds for data scientists. Older reports put the Life Story before pair programming; recent ML reports place it in the final loop. Team placement with a manager comes after the interviews, and several candidates describe long gaps between steps.
The Shopify interview process, stage by stage
01Recruiter or hiring-manager call
15-30 min video or phone call; some ML candidates go straight to the assessment or pair programming
A short conversation about your background, why Shopify and the role, plus logistics such as sponsorship and salary. The recruiter explains the later rounds and often tells you to prepare for the Life Story. One recent ML candidate was quizzed on basic facts about the company.
What it tests
- A clear, specific answer to why Shopify and why now
- A two-minute summary of your current role and best project
- Logistics: location, sponsorship, salary expectations
- Basic knowledge of what Shopify does and how it makes money
How to prepare
- Know the basics: what merchants use Shopify for, roughly when it was founded and how large it is.
- Ask which rounds follow, how long each is and whether AI tools are allowed in the coding round.
- Keep answers short; recruiters cut long scripted answers off.
An initial recruiter or manager call opens 30 ML and data-science reports; the company's FAQ describes it and says engineering applicants may skip to pair programming.
02Online assessment
Async, about 20-25 min of timed reasoning questions plus a long personality survey
A timed aptitude test with pattern, sequence, spot-the-difference and arithmetic questions, followed by a personality questionnaire that repeats similar items. Candidates rarely finish every reasoning question. Candidates are unsure how much it weighs in the decision.
What it tests
- Speed on number sequences and visual pattern questions
- Attention to detail under a tight clock
- Consistent answers across a long personality survey
How to prepare
- Do a timed practice set of cognitive-aptitude questions beforehand so the pace is not a surprise.
- Guess rather than leave items blank when time runs out; a calculator is allowed.
- Answer the personality survey honestly and consistently; it asks the same thing several ways.
Six reports from 2024-2026 (data-science and ML) describe an aptitude test, often with a personality survey; the candidate guide lists an online assessment before the craft round.
03Pair programming
About 75 min live: your own IDE, a GitHub repo to clone and push, tests expected; AI assistants allowed in recent ML reports. Data-science screens often split SQL and Python OOP
The craft gate. ML candidates now build a small, extensible program from a README or empty repo, such as a robot on a grid, a word game, a URL shortener or a refund engine, while the interviewer adds requirements and then asks about the code. Data-science screens more often hand you a buggy Python class (twice an A/B test helper) to explain, fix, test and extend, sometimes after a SQL half on merchant tables.
What it tests
- Clean object-oriented design that survives new requirements
- Writing unit tests without being asked
- Driving an AI assistant and explaining or correcting what it produces
- Debugging and extending unfamiliar code, and telling a code bug from a test bug
- Narrating trade-offs while you work with the interviewer
How to prepare
- Set up your IDE, Python environment, git and AI tool before the call; you will clone a repo and push at the end.
- Write tests early; one ML candidate was rejected for skipping them despite working code.
- Talk through every decision; reports say communication is graded as much as the result.
- For data-science screens, read the tables carefully and pick each join deliberately.
Described in 48 reports (12 ML, 36 data science); 2025-2026 ML reports agree on 75 minutes with a repo and tests, and AI tools from late 2025. The candidate guide names it.
04Life Story
About 60 min conversation with a recruiter
Shopify's signature round: you walk through your life and career in order while the recruiter stops you to ask why you made each choice, what was hard and what you learned. It covers education and personal turning points, not only jobs. Older loops ran it before pair programming; recent ML loops place it in the final round, which matches the company's note that it follows confirmation of your craft skills.
What it tests
- Why you made each transition, from your degree onwards
- Challenges you faced, decisions you owned and lessons you took
- Self-awareness and how you respond to setbacks
- Motivation for Shopify and fit with a fast, high-autonomy culture
How to prepare
- Prepare a chronological outline with the reason for every move; reports say interviewers dig into why you left each place.
- Keep each chapter short and let the recruiter steer; a session that ends well before the hour was read as a bad sign.
- Have one or two honest stories about failure and what changed afterwards.
23 reports (17 data science, 6 ML) describe an hour-long Life Story with a recruiter; placement moved from early to final loop in 2026 ML reports. The candidate guide describes it.
05Technical project deep dive
About 60 min on one or two of your own projects
You choose a project and defend it in detail: problem, data, model, trade-offs, results and your own part. ML interviewers probe classic ML knowledge along the way. One ML candidate was told their project read as data science rather than ML engineering, so pick work that shows engineering ownership for MLE roles.
What it tests
- Problem framing and business impact of your project
- Model and feature choices, and the alternatives you rejected
- Evaluation, deployment and what went wrong
- Classic ML fundamentals behind your choices
How to prepare
- Pick a project that matches the role: production ML for MLE, analysis and decisions for data science.
- Prepare numbers: data size, metrics before and after, latency or cost.
- Expect the same question asked several ways; answer at the level of detail asked.
15 reports (6 ML, 9 data science) include a project deep dive in the final loop, consistently about one hour; the candidate guide lists technical deep dives.
06ML system design (MLE)
About 60 min open discussion
Design an end-to-end ML system for a commerce problem: hierarchical product categorisation with overlapping tags, storefront search autocomplete, personalised product feeds or loan-offer recommendations. Interviewers often remove labelled data mid-round and ask how you would get labels, then push on real-time versus batch serving.
What it tests
- Scoping the product goal and success metrics
- Label strategy when no labelled data exists
- Candidate generation, ranking and serving architecture
- Real-time versus batch inference trade-offs
- Monitoring, cold start and iteration
How to prepare
- Have a plan for weak supervision, heuristics and human labelling ready; it came up in several reports.
- Practise multi-label and hierarchical classification, not just flat multi-class.
- Go deep on one component; the one detailed rejection reason was not enough depth here.
Free reported questions for this stage
- Personalized Product Feed for an Online StoreML System DesignHardSolve
Five ML-engineer reports from 2025-2026 describe ML system design rounds; one had two of them. Prompts vary but are all commerce problems.
07ML modeling & fundamentals (MLE)
About 60 min modeling discussion
A modeling-focused round on a business problem: data, features, model choice, training and evaluation, with less emphasis on infrastructure. Reported prompts are package delivery-time prediction, fraud detection on merchant transactions and product ranking. Earlier ML reports describe a separate ML-understanding round on classic topics such as regression and boosting.
What it tests
- Defining the target and when the prediction is made
- Feature engineering and leakage control
- Choosing and justifying a model family
- Offline metrics for imbalanced or ranking problems
- Classic ML: linear models, feature importance, boosting
How to prepare
- Structure every answer as data, features, model, training, evaluation; interviewers grade along those lines.
- For fraud, talk about class imbalance, cost-weighted thresholds and label delay.
- Brush up on traditional ML; one AI-engineer report says the questions favoured it over generative AI.
Five ML reports (2024-2026) describe a modeling or ML-understanding round; prompts were delivery ETA, fraud and product ranking.
08Data science case & analysis rounds (DS)
45-60 min each: product or marketplace case, SQL dataset exploration, or analysis in Python
Data-science finals replace the ML rounds with practical analysis: explore a dataset in SQL, interpret a chart, diagnose what drives profit, or design metrics for a new merchant feature. Some loops add a reverse case where you explain an analysis. Most of these reports are from 2021-2024, so confirm the current mix with your recruiter.
What it tests
- Metric design for a merchant or marketplace feature
- Turning a vague business question into an analysis plan
- SQL on realistic commerce tables
- Explaining a visualisation or result to a stakeholder
How to prepare
- Practise marketplace metrics: merchants and buyers, activation, retention, GMV.
- Say what decision your analysis would change before you start querying.
- Communicate the approach; reports say a clear process counted more than a perfect answer.
13 data-science reports describe case or analysis rounds, mostly 2021-2024; the candidate guide lists exploring a dataset in SQL. Not part of ML-engineer loops.
09Team placement & offer
About 30 min chat with a potential manager after the interviews
Shopify interviews you for a craft first and matches you to a team afterwards, so you may not meet your manager until the end. The hiring team reviews all feedback before deciding.
What it tests
- Mutual fit with a specific team
- Your questions about the work and the manager
How to prepare
- If meeting the team matters to you, ask early when placement happens.
- Ask for feedback after any decision; a few candidates got useful detail on request.
Two data-science reports mention team matching after the final loop; the candidate guide describes a 30-minute manager chat after interviews.
Questions and themes reported most often
- Walk through your life and career, explaining why you made each transition (23 reports, life story)
- Deep dive on one project you led, with detailed follow-ups (15 reports, technical project deep dive)
- Debug, test and extend a given Python class, sometimes an A/B test helper (8 reports, pair programming)
- Why Shopify, and basic facts about the company (8 reports, recruiter or hiring-manager call)
- Product sense: metrics for a new feature or what drives profit (6 reports, data science case & analysis rounds (ds))
- Build a small extensible app from an empty repo with tests: word game, URL shortener, refund engine (4 reports, pair programming)
- Robot on a grid: moves and turns, then obstacles, then many robots (3 reports, pair programming)
- Classic ML questions: linear regression, feature importance, boosting (3 reports, ml modeling & fundamentals (mle))
What Shopify asks about
- Pair programming & practical coding33%
- Life Story & behavioral18%
- ML system design15%
- ML modeling & fundamentals12%
- Project deep dive12%
- SQL & product analytics (DS)10%
Shopify interview prep checklist
- Built a grid-robot program with obstacles and multiple robots, plus tests, in under an hour
- Ran a full mock pair-programming session with your IDE, git push and AI tool
- Wrote a chronological Life Story outline with a reason for every move
- Prepared a one-hour deep dive on one project with real numbers
- Sketched a hierarchical product classifier and a storefront autocomplete end to end
- Can walk through a fraud or delivery-time model from target definition to evaluation
- Data science: fixed and extended a buggy Python class and wrote SQL over merchant tables
Tips
- Treat pair programming as the gate: practise building a small extensible program from an empty repo in 60 minutes, with tests, then explaining it.
- Rehearse with your own AI coding setup, and practise reading and correcting its output out loud.
- Write unit tests by default in every coding round, even when nobody asks.
- Prepare the Life Story like a structured interview: every transition needs a reason, a challenge and a lesson.
- For MLE finals, prepare one design and one modeling answer each for categorisation, search, recommendations and fraud.
- Have a ready plan for building a model without labels; interviewers return to it often.
- For data-science roles, drill object-oriented Python debugging alongside SQL; several strong analysts failed on the class exercise.
- Expect slow scheduling and chase your recruiter politely if a week passes without news.
See the rest of the Shopify loop
The later stages, the questions reported for them and how to prepare are in Pro.
Unlock with ProPractice the Shopify loop
Work the stages in order: every reported question for this company, grouped by round, with solutions in the app.
Shopify interview FAQ
How many interview rounds does Shopify have for ML roles?
Candidates report 9 stages: Recruiter or hiring-manager call, Online assessment, Pair programming, Life Story, Technical project deep dive, ML system design (MLE), ML modeling & fundamentals (MLE), Data science case & analysis rounds (DS) and Team placement & offer. A short recruiter or hiring-manager call (engineering-track candidates sometimes skip it), an online assessment, then a 75-minute pair-programming round that gates the final loop. The final loop is usually four interviews: Life Story, a project deep dive, ML system design and ML modeling for ML engineers, or SQL, analysis and product-case rounds for data scientists. Older reports put the Life Story before pair programming; recent ML reports place it in the final loop. Team placement with a manager comes after the interviews, and several candidates describe long gaps between steps.
How long does the Shopify ML interview process take?
Reported timelines run About 1-3 months.
What does Shopify ask in ML interviews?
By reported question mix: Pair programming & practical coding 33%; Life Story & behavioral 18%; ML system design 15%; ML modeling & fundamentals 12%; Project deep dive 12%; SQL & product analytics (DS) 10%. The most repeated themes: Walk through your life and career, explaining why you made each transition; Deep dive on one project you led, with detailed follow-ups; Debug, test and extend a given Python class, sometimes an A/B test helper.
How should I prepare for the Shopify ML interview?
Treat pair programming as the gate: practise building a small extensible program from an empty repo in 60 minutes, with tests, then explaining it. Rehearse with your own AI coding setup, and practise reading and correcting its output out loud. Write unit tests by default in every coding round, even when nobody asks.
Are there free Shopify interview questions?
Yes. LeetML has 15 questions reported from Shopify interviews, and 1 of them are free to solve in the browser; they are listed by stage above.
Related interview guides
- StripePractical multi-part coding, a timed dataset-to-model round, an ML bug squash and payments-flavoured ML design.
- eBayPractical Python and SQL coding, ML fundamentals and marketplace ML cases, with a heavy focus on your past projects.
- InstacartModerate coding, deep questions on classical ML and your own models, and grocery-flavoured ML design and experiment cases.
- WayfairA timed online assessment, then several business-flavoured ML case studies, one coding round and a behavioral round.
- DoorDashTimed SQL and marketplace case studies for data scientists; project deep dives, coding and infra-heavy ML design for ML engineers.
- AirbnbMarketplace ranking design, a hard look at your past models, one practical coding round and a values check.