EngMock

6 live AI-assisted interviews

Practice coding with AI under interview scrutiny

Work in a multi-file repository with EngMock’s own coding assistant, and show how you frame the task, direct the assistant, review and debug what it writes, and own the code you ship.

Start an AI coding mockBrowse scenarios

Key facts

  • EngMock has 6 AI-assisted coding interviews; two of them start from an existing codebase, where you fix defects, build a feature and then make it fast.
  • Every question can be answered in any of nine languages: Python, JavaScript, TypeScript, Java, C++, C#, Go, Rust and Kotlin.
  • The interview uses Runnable Coding: your code and tests run in an isolated sandbox, and the interviewer sees every run.
  • EngMock’s coding assistant reads the repository, plus the file you have open and the lines you select. It cannot see the interview question, the rubric or the interviewer, and it cannot edit or run your code.
  • EngMock’s assistant is the only AI that counts: AI used anywhere else is not visible to the interview and earns no credit.
  • Grading looks at your code, your runs, your conversation with EngMock’s assistant and what you explain to the interviewer. Assistant output is evidence of how you use the tool, never your own achievement.
  • In VS Code mode, the assistant opens in its own column to the right of your code.

What interviewers evaluate

AI output is visible. Your judgment is the signal.

01

Frame the task

Clarify acceptance criteria and inspect the existing files before asking the assistant to write code.

02

Delegate deliberately

Give the assistant bounded subtasks with context and constraints instead of handing over the whole problem.

03

Review critically

Read what comes back for correctness, security, maintainability and fit with the repository, then adapt or reject it.

04

Own the result

Run and test the final workspace and explain the code, including what the assistant wrote, as your own.

EngMock’s coding assistant

The assistant reads the repository you are working in and the file and lines you point it at. It does not receive the interview question, the rubric, your target level or the interviewer’s conversation, and it is read-only: it suggests code, and you decide what goes into the files. Your requests and its replies are recorded, so feedback can tell tool output apart from your review and implementation choices.

Live interviewer follow-ups

The interviewer sees a note of each request you send the assistant and asks about them at natural pauses: why you delegated, how you checked the answer, and what you would have done without it. It also probes requirement gaps, weak assumptions and your test strategy.

Existing-codebase tasks

Two questions start in someone else’s code

They run in three phases, and the tests are grouped by phase, so you always know what a run proves.

  1. 1

    Fix

    The existing code has defects. Find and fix them, and check more than the tests that already fail.

  2. 2

    Build

    Implement the next feature from the brief without breaking what already works.

  3. 3

    Optimize

    Switch on the large test tier, which stays skipped until then, and make the code fast enough to pass it.

The tool is allowed; uncritical delegation is not

Can I ask the assistant to write code?
Yes. The interview measures how well you scope the request, give it context, inspect the result and decide what to accept, adapt or reject. Asking it to solve the whole task, or pasting answers you cannot explain, are weak signals.
Can I use Copilot, ChatGPT or another AI tool instead?
Use EngMock’s assistant. The interview assesses your AI use from your conversation with it; AI used anywhere else is not visible to the interview and earns no credit.
What evidence is recorded?
Your requests to the assistant and its replies, your code changes, your runs and your spoken reasoning, so the report can separate generated output from your engineering judgment.
What does the interviewer probe?
Requirements, failure cases, security, maintainability, test strategy, missed defects, and whether you can explain and own the final implementation, including the parts the assistant wrote.

Related

Practice the AI-enabled coding round

Pick a task, use the assistant the way you would at work, and see how the report reads your judgment.

Start an AI coding mockBrowse AI Coding Question Bank