Why this conversation
The first five minutes decide the interview. Candidates who start drawing lose to candidates who scope, state assumptions out loud, and name the one constraint that shapes everything. This is that opening, sentence by sentence.
Dialogue
Interviewer: Design the settlement service for a prediction market. Take it wherever you like.
Jay: Before I draw anything, three questions, so I solve your problem and not one I invented. Is settlement on-chain, off-chain, or both? What's the scale, roughly, markets per day and positions per market? And what's the cost of settling wrong versus settling late?
Interviewer: Both, custody on-chain and matching off-chain. Say a thousand markets a day, ten thousand positions each. Wrong is much worse than late.
Jay: Then I'll say my key assumption out loud: correctness beats latency, so I'm designing a system that can stop. Every design choice from here follows from that. Push back if that's wrong.
Interviewer: That's fine. Go on.
Jay: Three components. A resolution source that produces a signed outcome. A settlement engine that computes payouts as a pure function of positions plus outcome, idempotent per market. And an invariant checker that compares total payouts against escrowed collateral before anything moves, and halts on mismatch.
Interviewer: Why the halt? Users will hate a frozen market.
Jay: They will, for an hour. You told me wrong is much worse than late, and a halt is the only reversible failure. If I'd been told the opposite, I'd design fail-open with a dispute window instead. That's the trade, and it's yours to make, not mine.
Interviewer: Good. Now let's talk about what happens at ten million positions.
Jay: Then the pure function is the thing I keep and the storage is the thing I change. Let me draw where it shards.
Key expressions
| Expression | 뜻 · 쓰이는 자리 |
|---|---|
| take it wherever you like | 원하는 방향으로 가져가라 |
| before I draw anything | 무엇을 그리기 전에 |
| a problem I invented | 내가 만들어낸 문제 |
| roughly | 대략 |
| wrong versus late | 잘못 대 늦음 |
| say my key assumption out loud | 핵심 가정을 소리 내어 말하다 |
| push back if that's wrong | 틀렸으면 반박하라 |
| a pure function of | ~의 순수 함수 |
| the only reversible failure | 되돌릴 수 있는 유일한 실패 |
| it's yours to make, not mine | 그 결정은 내 것이 아니라 당신 것이다 |