Skip to content
AI Engineer Guide.

concept · career · intermediate · concept 6 of 33

Mock interviews

0 of 33 complete

Why it matters

Interview performance is a skill separate from engineering ability, and it improves only with feedback under realistic pressure. Reading and re-reading notes builds familiarity with the material, not the delivery of it under observation. A mock compresses the loop: perform, observe, score, adjust. The value comes from the specificity of the feedback, which is why a rubric and a recording outperform a friend saying it went fine.

Mental model

It is a dress rehearsal with a stopwatch: the point is to find what breaks under stage conditions. The metaphor breaks because a rehearsal has a director, while a mock often has a peer who is reluctant to be blunt, so you must ask for the criticism.

How it works

The mock follows the weaknessInteractive
startquestionoutcomepath takenruled out
VerdictChoose· recommended hereConditional· with the named costStandard· the ordinary case

The rungs choose the instrument for the weakness; each stop names what it cannot show, and the last stop repeats the question after one change.

Step 1 of 6

Start with the weakness

The format follows the weakness you need to observe: delivery, level, basics or stamina. Pick one format and one question, and set the timer to the real round.

question

Is the round conversational?

A recording shows delivery but cannot interrupt you. If the real round is a conversation, the mock has to be one too.

Path walked

  1. You are about to book a mock
  2. Is the round conversational?
Step 1/6 · Start with the weakness

Figure. The mock is chosen by the weakness you need to see, not by how ready you feel. A recording shows delivery, a peer shows the basics, a professional shows the bar, and a full loop shows stamina; each stop names its instrument and what it cannot show, and the last stop repeats the question after one change.

What to notice. The order is the page's argument, not a measured ranking. No score threshold or price appears; “two mocks per format” is the page's own rule of thumb, not a measured count. The figure does not claim a professional mock predicts a pass: one mock is a sample of one interviewer, and the signal is the weakness it exposed.

Text version

A decision ladder for choosing a mock format. You are about to book a mock, and rung 1 asks whether the round is conversational; no leads to recording yourself, which answers delivery habits but cannot show interaction practice. Yes continues to rung 2: can a peer hold the level and be blunt? No leads to booking a professional mock, conditional on cost and availability; it answers the senior bar but cannot show a pattern. Yes continues to rung 3: are you still fixing basics? Yes leads to a peer mock with one fix at a time, cheap and schedulable; it answers basics under pressure but cannot show the level. No continues to rung 4: do you need stamina across a whole loop? Yes leads to a full loop simulation, conditional because it is expensive and premature before the basics hold; it answers stamina across a loop but cannot show depth on one answer. No leads to re-running the same question after one change; it answers whether the change held but cannot show a new behaviour, because two mocks per format is usually enough to find the pattern.

  1. Choose one format and one question, and set a timer that matches the real round.
  2. Ask the interviewer to stay in role and not coach you mid-answer, so the conditions resemble the real thing.
  3. Record the session, or at minimum take notes on the exact moment you lost structure.
  4. Score against a rubric covering structure, correctness, trade-offs and communication, immediately afterwards.
  5. Pick the single highest-impact weakness and write the replacement behaviour in one sentence.
  6. Re-run the same question a few days later and compare the scores, not the feeling.

Trade-offs

ChoiceGood forCostFails when
Peer mockCheap and easy to schedulePolite feedback and mismatched levelThe peer has never interviewed at that level
Professional mockRealistic and bluntCost and limited availabilityYou need many repetitions
Recording yourselfReveals delivery habitsNo interaction practiceThe round is conversational
Full loop simulationRealistic stamina and context switchingExpensive to arrangeYou are still fixing basics
Trade-off mapInteractive

Good for

Cheap and easy to schedule

Cost

Polite feedback and mismatched level

Fails when

The peer has never interviewed at that level

  1. The peer has never interviewed at that level
  2. You need many repetitions
  3. The round is conversational
  4. You are still fixing basics

Figure. Each choice with what it buys, what it costs, and the case where it loses.

What to notice. The fourth column is the decision: if the losing case describes your situation, the choice is wrong.

Text version

Four trade-offs: Peer mock is good for Cheap and easy to schedule, costs Polite feedback and mismatched level, and fails when The peer has never interviewed at that level. Professional mock is good for Realistic and blunt, costs Cost and limited availability, and fails when You need many repetitions. Recording yourself is good for Reveals delivery habits, costs No interaction practice, and fails when The round is conversational. Full loop simulation is good for Realistic stamina and context switching, costs Expensive to arrange, and fails when You are still fixing basics

Failure modes

  • Friendly feedback. The mock ends with encouragement and no specific change. Debug: ask the interviewer for one sentence describing what they would have written in the debrief.
  • No timer. Answers run long and the structure collapses under real time pressure. Debug: compare your answer length with the round duration.
  • Practising content only. You recall facts and still lose the thread when interrupted. Debug: have the interviewer interrupt at ninety seconds and note what happens.

In production

Keep a score sheet per format and review the trend before a real loop. Two mocks per format is usually enough to find the pattern; more without a change is repetition, not practice.

Common misconceptions

People think: Mocks are only useful when you are nearly ready.

Actually: They are most useful early, when a structural problem is cheap to fix and has time to become a habit.

People think: A good mock means you will pass.

Actually: One mock is a sample of one interviewer; the signal is the specific weakness it exposed, not the verdict.

Interview angle

When the conversation turns to improvement, a strong answer describes a measurement loop with a rubric, a recording and one change per iteration, and names a specific weakness a mock exposed. Weak answers describe mocks as confidence building and cannot say what they measured.

Sources