DeFi Protocol Interview Questions for Crypto Finance Jobs: I can explain AMMs, but I ramble when they ask LP risk, oracles, and what breaks

I’ve been on a few of these calls and the “explain AMMs” part is rarely the real test. The test is whether you can keep your answer tight and not drift into a mini-lecture.

What’s worked for me is a simple rhythm: mechanism → what can go wrong → what you’d check.

So I’ll say: “AMM is a pool and a pricing rule. Trades change the pool balance, price moves automatically. LPs earn fees for providing inventory.” Then I stop.

If it’s a protocol role, the follow-up is usually: “Okay, what’s the risk?” That’s where you pick one thread and go one level deeper: impermanent loss in trending markets, toxic flow/arbs taking the good side, or MEV making execution messy. If they lean in, then you bring in oracles and liquidation edge cases.

A small trick: end with “Want the intuitive version, or should I go into edge cases?” It shows control and keeps it conversational.

I’ve had the same problem — the questions sound simple, but they’re watching how you think more than the definition.

The shift for me was: stop trying to “teach DeFi” and answer like someone who’s thinking about a live protocol.

For DeFi vs TradFi I keep it short: “TradFi relies on institutions + processes. DeFi relies on contract logic + incentives.” If they want more, I talk about where trust moved (custody, execution, governance).

On AMMs/liquidity pools, I start with the picture: “pool + pricing rule.” Then I add one line that signals I’m not repeating Uniswap 101: “LP returns are fees minus adverse selection — arbs usually take the better side.” If they want deeper, I’ll go into IL, thin liquidity, and MEV in plain words.

For security, I don’t name-drop audits. I’ll mention one failure mode: oracle assumptions breaking, liquidation edge cases, or just bad parameters. That lands better because it sounds like real post-mortem thinking.

And if something is unfamiliar, I say it plainly: “I haven’t worked on your exact design, but here’s how I’d reason through it — docs, invariants, then oracle/liquidation/governance controls.” That honesty reads strong in these rounds.

I’ve seen this happen a lot in protocol interviews — you’re not actually failing on knowledge, you’re failing on answer shape.

Most interviewers aren’t waiting for a perfect AMM explanation. They’re watching whether you can do two things under pressure:

land the core idea fast, and

choose one risk thread and go one level deeper (instead of listing 10 risks).

What I started doing is answering AMMs like I’m talking to a teammate, not “teaching DeFi”:

“Pool + pricing rule. Trades move the balances, price updates. LPs earn fees for providing inventory.”

Then I pause. If they want depth, they’ll pull it out of you with a follow-up like “what’s the catch?” or “what breaks?”

When that happens, don’t try to cover everything. Pick one path and stay on it for 30–40 seconds. Example:

“LP returns are fees minus adverse selection. In a trending market you can underperform holding because arbs rebalance against you.” or

“If this protocol depends on an oracle for liquidations, the interesting question is what happens during volatility or thin liquidity.”

That’s usually enough for them to trust you. And if they do want math or edge cases, they’ll ask.

Also, on security/audits/regulation — I’d bring it up only if it changes a design decision. Otherwise it sounds like topic-dropping.

Curious: in your last 2–3 interviews, what was the exact follow-up after the “explain AMM” question — was it more about MEV, oracles/liquidations, or just “how would you design this differently?”