Skip to content
AI Engineer Guide.

drill · drills · beginner · concept 8 of 33

Behavioural: projects

0 of 33 complete

Why it matters

Project questions are a proxy for how you work when nobody is assigning tasks. The interviewer wants to see scoping, prioritisation, measurement and follow-through, because those are the parts of engineering that do not show up in a coding exercise. Candidates usually answer with a feature list, which reveals nothing about the choices behind it. The strongest answers are about a constraint that forced a decision.

Mental model

A project story is a case study: context, decision, intervention, result, reflection. The metaphor breaks because a published case study is edited for a reader and hides the dead ends, while an interview asks you to name them, including the ones that make you look uncertain.

How it works

Mechanism, step by stepInteractive
  1. 1Pick a project where you made a consequential decision and can still explain the alternatives you rejected.
  2. 2State the constraint first, such as a deadline, a legacy system or a fixed team, because constraints make the decision legible.
  3. 3Describe the option you chose and the one you did not, with the reason that separated them.
  4. 4Give the evidence of impact: a metric, an incident rate, a support load, or a decision that became possible.
  5. 5Say what you would change now, and what signal would have told you earlier.
  6. 6Keep the whole answer under two minutes, then let the interviewer choose where to dig.

Step 1 of 6

Pick a project where you made a consequential decision and can still explain the alternatives you rejected.

Step 1/6 · step 1

Figure. The mechanism in this page's own steps. The figure adds sequence, not new claims.

What to notice. A step can be read alone, but the order is the mechanism. The control lets you stop on any one.

Text version

The page's steps: 1. Pick a project where you made a consequential decision and can still explain the alternat… 2. State the constraint first, such as a deadline, a legacy system or a fixed team, because … 3. Describe the option you chose and the one you did not, with the reason that separated the… 4. Give the evidence of impact: a metric, an incident rate, a support load, or a decision th… 5. Say what you would change now, and what signal would have told you earlier. 6. Keep the whole answer under two minutes, then let the interviewer choose where to dig.

  1. Pick a project where you made a consequential decision and can still explain the alternatives you rejected.
  2. State the constraint first, such as a deadline, a legacy system or a fixed team, because constraints make the decision legible.
  3. Describe the option you chose and the one you did not, with the reason that separated them.
  4. Give the evidence of impact: a metric, an incident rate, a support load, or a decision that became possible.
  5. Say what you would change now, and what signal would have told you earlier.
  6. Keep the whole answer under two minutes, then let the interviewer choose where to dig.

Trade-offs

ChoiceGood forCostFails when
One deep projectShows judgement and depthNarrow coverage of your rangeThe role needs breadth across domains
Several short examplesShows rangeNo decision is examined closelyThe interviewer probes for depth and finds none
Emphasise the metricConcrete and verifiableCan hide the messy processThe metric moved for reasons outside your work
Emphasise the processShows how you workSounds like a methodology talkThe interviewer cannot tell whether it worked
Trade-off mapInteractive

Good for

Shows judgement and depth

Cost

Narrow coverage of your range

Fails when

The role needs breadth across domains

  1. The role needs breadth across domains
  2. The interviewer probes for depth and finds none
  3. The metric moved for reasons outside your work
  4. The interviewer cannot tell whether it worked

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: One deep project is good for Shows judgement and depth, costs Narrow coverage of your range, and fails when The role needs breadth across domains. Several short examples is good for Shows range, costs No decision is examined closely, and fails when The interviewer probes for depth and finds none. Emphasise the metric is good for Concrete and verifiable, costs Can hide the messy process, and fails when The metric moved for reasons outside your work. Emphasise the process is good for Shows how you work, costs Sounds like a methodology talk, and fails when The interviewer cannot tell whether it worked

Failure modes

  • Feature tour. The answer lists what was built and never names a choice. Debug: underline every sentence containing a decision and count them.
  • Unfalsifiable impact. The result is described as successful with no measurement. Debug: ask what number or behaviour would have shown failure.
  • No reflection. The story ends at launch and the interviewer has to prompt for lessons. Debug: check whether the answer contains a change you would make now.

In production

Write three project stories, one page each, with the decision, evidence and reflection. Rehearse the two-minute version and keep the details for follow-ups.

Common misconceptions

People think: The most impressive project is the biggest one.

Actually: A small project with a clear decision and measured result is easier to evaluate and usually scores better.

People think: You need a production launch to tell a good project story.

Actually: A cancelled project with a well-reasoned stop decision demonstrates judgement, which is what the question tests.

Interview angle

Strong answers establish the constraint, the alternatives, the decision and the evidence, then reflect without prompting. They distinguish what they owned from what the team did. Weak answers narrate a timeline of tasks and let the interviewer infer the judgement.

Sources