Meta's process is a phone screen with two coding problems, then an onsite with coding, system or product design (mid-level and above) and a behavioural interview. Since late 2025 one onsite coding round is AI-enabled.
What a mock looks like in the room, step by step. The panel, its order, the form and the criteria are the real round's; the words and numbers are illustrative.
The interviewer has your résumé open. Expect questions on anything you listed — every project and every tool — and a level pitched to the role you name at Meta.
In the order the real one moves: problem 1, then approach, then code, then test, then problem 2, then approach, then code, then test. Follow-ups come from what you just said.
Not run, as in most real screens: the interviewer reads it as you write and asks about your choices.
Scored on the 4 criteria this round is assessed on, with every line of your Résumé marked as held up, shaky or not yet asked. A practice estimate, never a forecast.
This board gives no marks of its own, so it is a practice index out of 100. For finding what to fix, not for predicting a result.
Nothing to buy first. Pick the round, fill in your Résumé, and the interviewer is ready. Your recording, if you ask for one, never leaves your device.
Face this interviewer yourselfOne engineer, on a video call with a shared editor.
Meta's coding interviews move fast — usually two problems in one interview — so speed with correctness matters as much as the idea.
There is no score. The screen decides whether you go to the onsite.
Assessed on: Problem solving · Coding · Verification · Communication
One engineer.
Since October 2025 one onsite coding round is AI-enabled instead — practise it as its own round. This is the classic one.
Assessed on: Problem solving · Coding · Verification · Communication
One engineer, with a shared editor, the code running, and an AI assistant beside it.
Meta introduced an AI-enabled coding round in October 2025: candidates fix a bug, build a feature and optimise it in an unfamiliar codebase, using an AI assistant — and are judged on whether they understand and verify what it writes.
Meta's real round runs in CoderPad with a file tree and a choice of models. Here the codebase is one file, and the assistant will not simply hand you the solution — reportedly neither does Meta's.
Assessed on: Problem solving · Code quality · Verification · Communication
One senior engineer.
For mid-level and above; a product-design variant is offered for some roles.
Assessed on: Requirements and scope · High-level design · Depth · Trade-offs and scale · Communication
One interviewer.
The behavioural interview helps decide the level of the offer as well as the offer itself.
Assessed on: Ownership · Collaboration · Handling ambiguity · Learning and growth · Communication
The interviewer has your résumé open. Expect questions on anything you listed — every project and every tool — and a level pitched to the role you name at Meta.
Every sitting — AI, real or peer — is scored on the same 4 criteria, so you can see a line move instead of weighing one opinion against another.
How it works, step by step: the mock interview guide, and coding, design and case rounds.
The interviewer reads your Résumé, asks in the real order and scores you on what the real one assesses. Free, and ready now.
