#Behavioural Questions and the STAR Bank

Databricks publishes six cultural principles and interviews against them closely. Culture fit is weighted in the decision, not a formality. Even in a non-Databricks loop, these six are a good scaffold because they cover what senior interviewers actually probe.

We are customer obsessed · We raise the bar · We are truth seeking · We operate from first principles · We bias for action · We put the company first

#The format: STAR-L

Situation (2 sentences) → Task (1–2) → Action (the bulk, and say "I") → Result (with a number) → Learning (honest).

Three failure modes to avoid:

• The "we" problem. Senior candidates describe team achievements and never their own decisions. Interviewers write down "unclear what they personally did." Say I decided, I pushed back, I chose — credit the team separately and explicitly.
• No number. "It went well" is not a result. Estimate if you must, and flag it as an estimate.
• No learning. Every story that ends in unblemished triumph reads as unreflective.

Length: three minutes. Time yourself. Most people run to six and lose the room at four.

#Build your six

Fill these in properly — a half-prepared story is worse than an unprepared one, because you commit to it mid-answer and then discover it doesn't support the point.

#1. Customer obsessed

Prompt: "Tell me about a time you advocated for the customer against internal pressure."

Strong shapes: you told a customer not to buy something · you rebuilt something that technically met spec but didn't solve the real problem · you absorbed short-term pain for their long-term outcome.

#2. We raise the bar

Prompt: "Tell me about a time you raised the standard of work around you."

Strong shapes: you introduced a practice that outlived you (review standards, a paved road, an eval harness) · you rejected work that met the letter of the requirement · you rewrote something "finished" because it wasn't good enough.

#3. We are truth seeking

Prompt: "Tell me about a time you were wrong." / "...when the data contradicted what leadership believed."

This is the most important one and the most commonly fumbled. Strong shapes: you ran the benchmark that disproved your own recommendation · you told an executive their premise was wrong, with evidence · you reversed a decision you'd championed publicly.

Pick a real failure with real cost. Not "I worked too hard."

#4. First principles

Prompt: "Tell me about a time you solved a problem by going back to fundamentals."

Strong shapes: everyone assumed a constraint that turned out not to exist · you diagnosed something from mechanism rather than pattern-matching · you rejected the standard answer because the standard answer's assumptions didn't hold here.

#5. Bias for action

Prompt: "Tell me about a time you moved without complete information."

Strong shapes: you shipped something imperfect to unblock a decision · you made a reversible call quickly rather than a perfect one slowly · you broke an analysis-paralysis deadlock.

Show the judgement, not just the speed — say why the decision was reversible or the risk bounded. Bias for action without judgement reads as recklessness.

#6. Company first

Prompt: "Tell me about a time you did something that helped the wider organisation at your own or your team's expense."

Strong shapes: you gave up headcount, scope or credit for a better overall outcome · you killed your own project · you supported a decision you'd argued against, properly rather than grudgingly.

#Twenty-four questions to be ready for

Leadership and influence

1. Tell me about a time you influenced a decision without authority.
2. Describe a technical decision you lost. What did you do next?
3. How do you handle an engineer who disagrees with your architecture?
4. Tell me about the most difficult stakeholder you've worked with.
5. How do you get a room of opinionated architects to converge?
6. Describe a time you had to say no to a senior executive.

Delivery 7. Tell me about a programme that slipped. What did you do? 8. Describe the largest migration you've led. What went wrong? 9. How do you sequence a multi-year platform programme to keep support? 10. Tell me about a time you cut scope. Who was unhappy and how did you handle it? 11. Describe an incident you led the response to. 12. How do you handle being brought in to rescue something failing?

Technical judgement 13. What's the worst architecture decision you've made? 14. Tell me about a time you chose the less exciting technology. 15. Describe a time you were the only one who saw a risk. 16. When did you last change your mind about a technology you advocated? 17. Tell me about something you built that you'd build differently now.

Customer and commercial 18. Tell me about a customer relationship you turned around. 19. Describe a time you lost a deal or a customer's confidence. Why? 20. How do you handle a customer who wants something you think is wrong? 21. Tell me about explaining something deeply technical to a non-technical executive.

Self 22. What are you worse at than people assume? 23. What's the most useful feedback you've had in the last two years? 24. How do you stay current? (Be specific and verifiable — and this file's sibling ../01-platform/09-platform-2026-whats-new.md should make your answer easy to evidence.)

#Question 22 deserves preparation

"What are you worse at than people assume?" and its cousin "what's your weakness?"

Bad: a strength in disguise ("I'm too detail-oriented"). Bad: something disqualifying ("I don't really do stakeholder management"). Good: a real, bounded weakness with evidence you are managing it.

"I'm slower than I should be to delegate the technical work I find interesting. On the last programme I kept the hardest performance problem for myself for about three weeks when I should have paired someone into it, and that made me the bottleneck. What I do now is name the piece I want to keep at the start, and deliberately give it away."

That is specific, plausible, and shows self-correction — which is the actual thing being tested.

#Rehearsal protocol

1. Write all six stories out in full. Yes, write them — the discipline exposes the ones with no number and no learning.
2. Record each. Listen. Cut to three minutes.
3. Re-record until you can tell each without notes, in three minutes, with the number in the result and a genuine learning.
4. Then practise re-aiming: the same story answers several questions depending on emphasis. Your migration story can serve influence without authority, a programme that slipped, or first principles — what changes is which part gets the airtime. Practise that pivot; it is what stops you panicking when the question is not one you prepared.