In this article
Before you start
Everything else, including the scoring and the round names, is explained below.
Why interviewers ask this
Meta asks these questions because an engineering manager's output is other people's work, and the only evidence a loop can collect about that is what you did in the past and how you reasoned about it. The consequence for you is that a polished philosophy with no example earns little: each round is a set of prompts that need a specific story with a decision and an outcome in it.
What the loop separates is a manager who can point to decisions and results from one who describes good intentions. A weak answer says "I believe in empowering my team". A strong answer says what you did when a high performer was unhappy, what you changed and what happened next. A third-party guide reports that the recruiter screen is where your level is set, and that level is the bar for the rest of the loop (Prepfully, updated September 2026).
The Meta engineering manager interview from first principles
In plain words
- Engineering manager
- An engineering manager is a person who leads a team of software engineers and is responsible for the team's delivery, its people and its technical health, rather than for writing most of the code personally.
- Project retrospective
- A project retrospective is a structured look back at one finished project in which you explain each major decision, what you knew when you made it and what the outcome taught you.
- Cross-functional partner
- A cross-functional partner is a colleague from another discipline, such as product or data science, whom you work with on shared goals but do not manage.
Think of the loop as a flight check for a pilot. The pilot flies a simulator. The pilot answers questions about weather. The pilot talks through a past emergency. Together the checks show whether the pilot can be trusted to fly. The analogy breaks in one place. A pilot's checks are fixed, but a Meta loop varies by level and over time. So prepare building blocks and confirm the format with your recruiter.
In one precise sentence: you pass a recruiter call, a screening and a loop of about a day, where most rounds score how you lead people, plus a design round and an AI-enabled coding round.
Step-by-step description
- A recruiter call, a screening, then a full loop.
- Five signals score the people rounds.
- Each story covers at least two signals.
- Prepare one project you can defend decision by decision.
- Practise design aloud, then the AI-enabled coding round.
- Meta says the process typically has a recruiter conversation, an initial screening and a full loop.
- A third-party guide says the four non-technical rounds score five signals and each story is scored against at least two.
Step 1: Learn the stages of the process
The first step is to learn the order of the stages, because each stage tests something different. Meta's hiring page says the first step for most roles is a call with a recruiter about your background, experience, long-term goals and interest in Meta, and that the process typically includes that conversation, an initial screening and a full loop interview with Meta employees (Meta hiring process page, read October 2026). It says a process takes about two to three months. The input is your resume and the output is the loop invitation.
A third-party guide describes a 30-minute recruiter screen, a 45-minute initial interview covering people management, a piece of technical design and career motivation, then an onsite of five to seven rounds, most of them 45 minutes, usually in one day (Prepfully, updated September 2026). The guide adds a hiring committee and team match step after the loop. Treat these counts as a guide; they come from a third-party guide, not a Meta publication. What you now know: the order of the stages, and that the people material comes first.
Step 2: Map each round to the five signals
The second step is to learn what each round scores, because every story is scored against at least two of the five signals, and the same signals recur across the people rounds. The Prepfully guide lists the rounds as people management and cross-functional collaboration, a project retrospective, cross-functional partnerships, building management and engineering culture (more common at the higher manager grade), a technical project deep dive, system or product design, and an AI-enabled coding round. It says the four non-technical rounds score five signals: driving results, embracing ambiguity, resolving conflicts, growing continuously and communicating effectively, and that each story is scored against at least two of them.
The common error is preparing by round, with one story for each, instead of by signal. The input is a round name, the change is a mapping to the signals, and the output is a list of stories you need. The signals are the guide's reading, not a Meta publication; confirm with your recruiter. What you now know: you need a bank of stories that covers all five signals, with spares.
Step 3: Build a story bank and check coverage
The third step is to build a story bank. A story bank is a small set of true stories, each outlined with the signals it can honestly evidence, and you check that every signal has a story. Meta's advice to engineering candidates (a 2022 post written for software-engineer interviews) is to use simple language when giving your answers (Meta Careers, updated October 2022).
The input is your last few years of management work, the change is tagging each story, and the output is a grid. Typical prompts that need a story include handling an underperforming engineer, managing a high performer, changing direction midway and settling a dispute between product and engineering. The worked example below runs the check on an invented bank. What you now know: coverage is a property of the whole bank, so look at the grid and not one story at a time. The wider method is in the behavioral interview guide.
Step 4: Prepare the project retrospective
The fourth step is to choose one project you can defend decision by decision. The Prepfully guide says the retrospective tests execution judgment: the interviewer probes each decision against its outcome, with prompts such as changing direction midway or cutting a launch's scope to test a hypothesis. It adds that the technical project deep dive is run as a live dialogue by a senior engineer and that slides are discouraged. The guide adds that candidates often get only one of the retrospective and the deep dive, so choose a project that serves both.
The input is one project, the change is writing its timeline with the decision and the information you had at each point, and the output is a story that survives follow-ups. Write down what you would do differently. A project that went perfectly leaves nothing to discuss. What you now know: the retrospective rewards honest reasoning over a clean result.
Step 5: Prepare design and the AI-enabled coding round
The fifth step is to prepare the design and coding rounds. Meta's hiring page says many technical interviews include a built-in AI assistant that candidates are expected to use, that design interviews run in CoderPad (an online shared coding editor) with Mermaid Markdown (text that renders as a diagram) using the same assistant, and that outside AI tools are not permitted (Meta hiring process page, read October 2026). The Prepfully guide says the AI-enabled coding round is the only coding round at the first manager grade, that it is 60 minutes on a multi-file codebase, and that it scores problem solving, code understanding, verification and debugging, and technical communication.
The input is a prompt and the output is working, verified code. Meta's 2022 engineer post advises writing actual code, not pseudocode. For design, see Meta's product architecture round and system design. What you now know: the technical rounds test how you work with the tool and explain trade-offs, not only whether you can code.
Knowing it is half the job. Saying it is the other half.
Interview Clozo is an AI interview copilot for real interview rounds: it hears the question and puts a strong answer on your screen, and you say it in your own words. See the features or how it works.
Start Free Trial →Worked example
The five stories below are invented to show the method; yours come from your own work. Each is tagged with the signals it can honestly evidence, and a short script finds the smallest set of stories that covers all five.1
| Story | What it shows | Signals it evidences |
|---|---|---|
| A | Cut a launch to test a hypothesis | Driving results, embracing ambiguity |
| B | Settled a product and engineering dispute | Resolving conflicts, communicating effectively |
| C | Coached a mid-level engineer to senior | Growing continuously, communicating effectively |
| D | Fixed an on-call burden (the pager rota for out-of-hours incidents) | Driving results, growing continuously |
| E | Changed direction midway | Embracing ambiguity, driving results |
Three stories are enough to cover all five signals, and four sets work: A with B and C, A with B and D, B with C and E, and B with D and E.1 Driving results has three stories, four signals have two or fewer, and resolving conflicts has only story B. If two rounds both ask a conflict question, story B is used twice, so the bank needs a second conflict story. Three or four 45-minute people rounds are 135 to 180 minutes of interview time, each with several prompts, so a five-story bank will be reused; plan at least two stories for every signal.1
Text description of the diagram
- In an illustrative five-story bank, four sets of three stories cover all five signals.
- Resolving conflicts has one story, and driving results has three.
Trade-offs and edge cases
The structure above mixes Meta's own pages with a third-party guide, so it describes how candidates report the process, not a Meta rulebook. Three limits follow.
First, the loop varies by level and over time. The Prepfully guide says the first manager grade (M1) is where most external hires land, that the next grade (M2) is often manager-of-managers, and that the culture round is more common at the higher grade. Ask your recruiter which rounds apply to you. Second, the guide itself gives both five to six and five to seven rounds, and Meta's page gives no round count for engineering managers, so treat any list as a sample. Third, the AI rules change: Meta's page says select roles include the assistant, so check the current rule for your role.
| Stage or round | Length | What it tests (reported) | Source |
|---|---|---|---|
| Recruiter screen | 30 min | Background, goals, interest in Meta; level set | Meta hiring page and Prepfully, September 2026 |
| Initial interview | 45 min | People management, technical design, career motivation | Prepfully, September 2026 |
| People management and cross-functional collaboration | 45 min | Managing engineers and working with stakeholders | Prepfully, September 2026 |
| Project retrospective | 45 min | Execution judgment, decisions against outcomes | Prepfully, September 2026 |
| Cross-functional partnerships | 45 min | Influencing teams you do not manage | Prepfully, September 2026 |
| Building management and engineering culture | 45 min | Team culture (more common at M2) | Prepfully, September 2026 |
| Technical project deep dive | 45 min | A past technical project, run as a live dialogue | Prepfully, September 2026 |
| System or product design | 45 min | Trade-offs at Meta scale | Prepfully, September 2026 |
| AI-enabled coding | 60 min | Debugging and verifying a multi-file codebase | Prepfully, September 2026 |
| Hiring committee and team match | not reported | Decision and placement after the loop | Prepfully, September 2026 |
For another company's published values, compare Google behavioral interview questions.
Text description of the diagram
- A third-party guide says every story is scored against at least two of five signals.
- Resolving conflicts has only story B.
How to answer it in an interview
Tell me about a time you handled an underperforming engineer.
One of my senior engineers was missing commitments for two quarters. My goal was to find out why before judging. In a one-to-one I learned that on-call load was eating his week, so I moved two services to another rotation and set three written four-week goals with weekly check-ins. He met two of three goals, missed the third and we agreed a clear next step. What I changed in my own habits is that I now check workload before performance, and I would say the earlier I raised the problem, the better it went.
The spoken structure is: goal and reason, what you did, the outcome including what did not work, and what you do differently now. It is about 95 words, sayable in under a minute, and it names one decision the interviewer can probe.
Common mistakes
- Telling a general success. (wrong) "We improved delivery and the team was happier." (right) "I moved two services off one rotation, and he met two of three goals." Why it is wrong: the interviewer cannot probe a decision that is not named.
- Preparing one story per round. (wrong) "I have a story for each round." (right) "I have stories tagged by signal, with a spare for conflict." Why it is wrong: a one-per-round bank never checks coverage, and each round asks several prompts, so a signal can end up with one story or none.
- Hiding the failure. (wrong) "Everything went as planned." (right) "Two of three goals were met; here is what I changed." Why it is wrong: follow-up questions find the gap, and hiding it costs trust.
- Ignoring the technical rounds. (wrong) "Managers do not code." (right) "I have practised the coding and design rounds aloud with the assistant." Why it is wrong: the loop includes design and AI-enabled coding rounds.
- Reading slides. (wrong) "Let me share my deck." (right) "I will talk it through and sketch the architecture." Why it is wrong: the guide says the deep dive is a live dialogue and slides are discouraged.
Follow-up questions
What are you doing now, what is exciting about it, and why Meta?
Answer in three short parts: your current scope and one result, the part of the work that energises you, and one reason tied to the team's problem. A third-party guide reports that the initial interview also covers career motivation (Prepfully, updated September 2026), so prepare this answer early and keep it short.
How do you handle performance in your team, both high performers and low performers?
Say what you do for each: for a high performer, a growth plan and more scope; for a low performer, a clear expectation, a written plan, regular check-ins and an honest end point. Back each with a story, as in the answer above. More on management stories is in engineering management.
Is there a coding round for engineering managers at Meta?
Yes, according to the Prepfully guide: one 60-minute AI-enabled coding round, which it calls the only coding round at the first manager grade. Meta's page says many technical interviews include a built-in assistant and candidates are expected to use it.
What is the difference between M1 and M2?
The Prepfully guide describes M1 as where most external manager hires land and M2 as often manager-of-managers, with the culture round more common at M2. Meta's pages give no grade definitions, so confirm with your recruiter.
What if the loop has only three people rounds?
Then you have 3 x 45 = 135 minutes of interview time instead of 180, and the same five signals still score every round. The Prepfully guide says the fourth people round is more common at M2, so an M1 loop may have three. Keep at least two stories per signal either way.
Can you use AI tools in the Meta interview?
Meta's page says select roles include an authorized assistant in CoderPad, and that outside AI tools are not permitted. Practise with the assistant in the practice session the page says is available through your Career Profile, and ask your recruiter what applies to your role. For your own questions at the end see questions to ask the interviewer.
- Meta describes a recruiter call, an initial screening and a full loop of interviews.
- A third-party guide says four non-technical rounds score five signals, and every story is scored against at least two.
- Keep at least two stories for every signal, so a second conflict question does not reuse the same story.
- Choose one project you can defend decision by decision, and do not rely on slides.
- Practise the AI-enabled coding and design rounds, and confirm the current format with your recruiter.
Sources
- Meta Jobs hiring process — Meta (page shows no date; read 2026-10-25), 25 Oct 2026
- Interviewing at Meta: the keys to success — Meta Careers (originally published 2020-08-25; updated 2022-10-14), 14 Oct 2022
- Meta Engineering Manager Interview Guide — Prepfully (third-party; last updated 29 September 2026), 29 Sep 2026
