Company interview guide

McKinsey Machine Learning Interview Guide (2026)

Data science interviews wrapped in a consulting loop: online test, project deep dives, client-style cases and probing personal-experience interviews.

McKinsey's machine learning interview loop has 6 reported stages (Recruiter call, Online assessment, Technical experience interview, Case interviews, Personal Experience Interview, Final round); typical length: Several weeks.

Roles covered: Data Scientist, ML Engineer and AI Engineer. The question bank holds 2 questions reported from McKinsey interviews, 1 of them free to solve in your browser.

Try a free question

Interview process length
Several weeks
Stages
6
Reports linked to bank questions
2
Bank report dates
Dec 2025–Sep 2026
Questions in bank
2 (1 free)
Last reviewed
October 2026

How the McKinsey interview loop works

A recruiter or HR call and a timed online assessment come first, in either order. Candidates who pass usually get a first round of two interviews (a technical experience interview plus a PEI question, and a data science or analytical case), then a final round of two or three interviews, often in person and led by partners, mixing business cases with more PEI. Some offices add a live coding or pair-coding session, and a few teams set a take-home that is discussed in a later interview. Several candidates describe feedback or coaching from the recruiter between rounds. Reports rarely give an overall timeline; a few call the process long.

The McKinsey interview process, stage by stage

01Recruiter call

Short phone or video call with HR or a recruiter, about 30 min

A screen on your background, motivation and logistics. Expect why McKinsey or why QuantumBlack, a walk through your resume and sometimes a question on your most impressive project or whether you are comfortable travelling for client work. In some processes this call comes after the online assessment rather than before it.

What it tests

  • Why McKinsey or QuantumBlack, and why a consulting setting
  • A concise walk through your resume and best project
  • Comfort with travel and client-facing work
  • Logistics: office, level, timeline

How to prepare

  • Have a specific answer to why QuantumBlack that goes beyond the brand.
  • Ask the recruiter for the Candidate Preparation Booklet and the format of each round.
  • Prepare a two-minute version of your most relevant technical project.

An HR or recruiter screen appears in 16 data-science reports, 14 of them since 2024; three put it after the online test. The company's page only says recruiters share role-specific preparation.

02Online assessment

Timed online test, from about 30 min up to 2 hours; contents vary by office

A hands-on screen that often cuts the funnel sharply. Reported versions include two Python and two SQL questions, multiple-choice sections on statistics and ML theory, algorithmic coding (one report had BFS for minimum knight moves, another two hard optimisation puzzles with hidden tests) and a timed end-to-end modeling task: load data, train, predict and save results to a CSV against an error threshold. Several reports used two separate tests, one for coding and one for statistics or ML.

What it tests

  • Python coding, from data manipulation to graph search
  • SQL queries
  • Statistics and ML theory in multiple-choice form, including regression and regularization
  • Building a working model quickly on a provided dataset

How to prepare

  • Budget time per section; one candidate ran out of time for the modeling task after the coding puzzles.
  • For the modeling task, get a simple baseline saved in the right output format first, then improve it.
  • Test edge cases yourself, since only some test cases are visible.
  • Review regression, L1 / L2 regularization and textbook ML definitions for the quiz section.

Free reported questions for this stage

An online test appears in 18 reports, 16 since 2024; content differs by office (coding, SQL, ML quiz, modeling task). The company's page says technical roles may take a hands-on technical assessment.

03Technical experience interview

Up to about 60 min with a senior data scientist or manager; sometimes a 20-25 min segment next to a PEI

McKinsey's technical experience interview (TEI) starts from a project on your resume and digs into data, modeling choices, validation and impact, then widens into ML and statistics fundamentals. Reported questions cover decision trees versus random forests, boosting, ensembles, bias and variance, overfitting in neural networks, k-means and how to choose k, interpreting linear regression output, and deep learning basics such as CNNs, RNNs and transformers. Some offices add a short coding exercise or pair-coding session, SQL, or an architecture review; a few teams set a take-home and spend this interview on your solution.

What it tests

  • Explaining one project end to end: problem, data, model, validation, impact
  • Classical ML fundamentals: trees, ensembles, bias-variance, regularization, clustering
  • Reading and interpreting regression output
  • Explaining a model and its limits to a non-technical client
  • Occasionally live coding, SQL or a review of a system you built

How to prepare

  • Pick one or two projects you can defend at every level, including what you would do differently.
  • Be ready to explain how you validated a model and how you would justify it to a sceptical executive.
  • Revise the formulas and intuition behind standard models; some interviewers ask how an algorithm works step by step.

A technical or project deep-dive interview appears in 21 reports, 17 since 2024; three mention live or pair coding and two a take-home. The company's page lists technical and expertise interviews.

04Case interviews

Up to about an hour each; usually one data science case early and one to three business cases later

Interviewer-led problem-solving cases built on a client scenario. Early cases often have a data or ML angle: which factors drive an outcome, which regression coefficients to look at, how to design a model or an experiment for a client. Later cases are often the same business cases consultants get, such as improving customer intake for a retailer or launching a new product in a region. Candidates stress that you must drive the structure yourself, ask for data and build a clear breakdown before naming any model; one candidate was turned down after the final round for letting the interviewer drive the case.

What it tests

  • Structuring an ambiguous client problem into clear, non-overlapping parts
  • Choosing and justifying an analytical or ML approach for a business question
  • Quick quantitative reasoning and interpreting data you are given
  • Turning findings into a recommendation for the client

How to prepare

  • Practise standard consulting cases, not just data science cases; the final round may have no technical content.
  • State the business goal and a structure before suggesting a model.
  • Drive the case: propose next steps rather than waiting for the interviewer to lead.
  • Work through the practice cases on the company's interviewing page.

Case interviews appear in 20 reports, 17 since 2024; about half describe a data or ML case, the rest classic business cases. The company's page describes the problem-solving interview.

05Personal Experience Interview

Often a 15-20 min segment at the start of a case or technical interview; asked in most rounds

McKinsey's PEI asks for one specific past experience and then probes it in depth: what you did, why, what others did and what you would change. Reported prompts include disagreeing with your manager, convincing senior people, having personal impact, leading through conflict and handling a hard project. Candidates call it deeper than a usual behavioral interview, and it repeats across rounds, so you need several distinct stories.

What it tests

  • Personal impact: a moment where your actions changed an outcome
  • Influencing or disagreeing with more senior people
  • Handling conflict or a difficult project
  • Clear, honest reflection under detailed follow-up questions

How to prepare

  • Prepare at least three distinct stories, since different interviewers may ask the same theme.
  • Focus on your own decisions and reasoning, not the team's.
  • Expect follow-ups on trade-offs and alternatives; keep details you can expand on.

PEI or behavioral questions appear in 17 reports, 12 since 2024, mostly paired with a case or technical interview. The company's page describes the PEI.

06Final round

Two or three back-to-back interviews, often in person and often with partners

The last round usually repeats the case and PEI formats with more senior interviewers, sometimes with one more technical interview. Candidates describe it as less structured than earlier rounds; one partner set a probability question, another round centred on a presentation and fit. One candidate's feedback after this round was that they needed to drive the case thinking themselves.

What it tests

  • Business cases with senior interviewers
  • More PEI stories and fit with the firm
  • Sometimes a final technical interview or presentation

How to prepare

  • Use the recruiter's feedback from earlier rounds; several candidates got specific coaching notes.
  • Expect less structure and adapt to the interviewer's style.
  • Keep fresh PEI stories in reserve for this round.

A distinct final or second round appears in 10 reports, nine since 2024; five say it was in person or with partners. The company's page does not describe round structure for technical roles.

Questions and themes reported most often

  • Deep dive on your most relevant project, then how you validated it and would explain it to a client (9 reports, technical experience interview)
  • Disagreeing with your manager or convincing senior people (4 reports, personal experience interview)
  • Python and SQL questions in the online test (4 reports, online assessment)
  • Why McKinsey or QuantumBlack, and are you comfortable travelling (3 reports, recruiter call)
  • Decision trees vs random forests, boosting and ensembles (3 reports, technical experience interview)
  • Linear regression: what it is, reading its output and which coefficients matter (3 reports, technical experience interview)
  • Bias-variance, overfitting and L1 / L2 regularization (3 reports, technical experience interview)
  • Timed modeling task: load data, train, predict and export results (2 reports, online assessment)

What McKinsey asks about

  • Case interviews30%
  • Personal experience (PEI)22%
  • ML & stats concepts20%
  • Project deep dive15%
  • Coding & SQL13%

McKinsey interview prep checklist

  • Finished a timed online-test mock with Python, SQL and a small modeling task
  • Can explain trees, random forests, boosting and regularization without notes
  • Can interpret a linear regression table and say which coefficients matter
  • Have a project deep dive covering data, model choice, validation and business impact
  • Prepared at least three PEI stories with distinct themes
  • Worked through several business and data science cases, leading the structure myself
  • Have a specific answer to why McKinsey or QuantumBlack

Tips

  • Prepare for the cases and PEIs as seriously as the technical rounds; they decide many outcomes.
  • Practise a timed modeling task end to end, including saving predictions in the required format.
  • Drill Python and SQL basics plus a few BFS and array problems for the online test.
  • Revise classical ML theory: trees and ensembles, regularization, bias-variance, clustering and regression output.
  • Choose one or two projects and rehearse explaining them to both a technical lead and a sceptical executive.
  • Write out three or more PEI stories on impact, influence and conflict, each with your own decisions at the centre.
  • Practise interviewer-led cases where you drive the structure, using the firm's published practice cases.
  • Ask your recruiter for the preparation booklet and act on any feedback you get between rounds.

See the rest of the McKinsey loop

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

Unlock with Pro

Practice the McKinsey loop

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

McKinsey interview FAQ

How many interview rounds does McKinsey have for ML roles?

Candidates report 6 stages: Recruiter call, Online assessment, Technical experience interview, Case interviews, Personal Experience Interview and Final round. A recruiter or HR call and a timed online assessment come first, in either order. Candidates who pass usually get a first round of two interviews (a technical experience interview plus a PEI question, and a data science or analytical case), then a final round of two or three interviews, often in person and led by partners, mixing business cases with more PEI. Some offices add a live coding or pair-coding session, and a few teams set a take-home that is discussed in a later interview. Several candidates describe feedback or coaching from the recruiter between rounds. Reports rarely give an overall timeline; a few call the process long.

How long does the McKinsey ML interview process take?

Reported timelines run Several weeks.

What does McKinsey ask in ML interviews?

By reported question mix: Case interviews 30%; Personal experience (PEI) 22%; ML & stats concepts 20%; Project deep dive 15%; Coding & SQL 13%. The most repeated themes: Deep dive on your most relevant project, then how you validated it and would explain it to a client; Disagreeing with your manager or convincing senior people; Python and SQL questions in the online test.

How should I prepare for the McKinsey ML interview?

Prepare for the cases and PEIs as seriously as the technical rounds; they decide many outcomes. Practise a timed modeling task end to end, including saving predictions in the required format. Drill Python and SQL basics plus a few BFS and array problems for the online test.

Are there free McKinsey interview questions?

Yes. LeetML has 2 questions reported from McKinsey interviews, and 1 of them are free to solve in the browser; they are listed by stage above.

Related interview guides

Official sources