Interview Tactics Playbook

You know the technical material cold. You walk into the interview anyway — and ramble for 20 minutes about load balancers, never reach the deep dive, get flagged as "unfocused" in the debrief. Knowing system design is half the battle. The other half is performing it under time pressure, navigating ambiguity, knowing when to push back, knowing what to drop.

This page is the operating manual for the 60-minute window: what to say in the first 5 minutes, how to time-box, when to admit "I don't know," what scoring rubric interviewers actually use. The technical pages teach the material. This page teaches how to deliver it.

The single most-failed phase. Candidates jump to architecture; interviewers think "no rigor."

What to actually do:

1. Restate the problem in your own words. "So you want me to design Twitter — specifically the home feed reading and posting flows. Is that right?" Confirms scope.
2. Ask about scale. "Roughly how many users are we designing for? 1M, 100M, 1B?" The answer changes everything.
3. Ask about features. "Should I focus on posting + reading, or also include search, DMs, ads?" Most interviewers want you to scope down.
4. State assumptions out loud. "I'll assume 100M DAU, mostly mobile, read-heavy 100:1, eventual consistency on feeds is OK." Get them to nod or correct.

By minute 5, you've earned credibility and bounded the problem. Now you can build.

Successful candidates run an internal clock and announce phase transitions. Approximate breakdown for a 60-minute interview:

Say out loud "I have about 30 minutes left, let me move to the deep dive." Interviewers love this — it shows self-awareness. If they want you to spend longer somewhere, they'll redirect.

The interviewer asks: "How does Cassandra handle anti-entropy at the partition level when ring rebalancing happens mid-repair?" You don't actually know.

Three responses, ranked:

1. Worst: bluff. "Yeah so it uses a Merkle tree thing and it just kind of works it out..." — interviewer instantly notices. Hard signal: this candidate fakes when uncertain.
2. OK: surrender. "I don't know that off the top of my head." — neutral but flat. No partial credit.
3. Best: think out loud. "I don't know the exact mechanism, but here's how I'd reason about it: Cassandra uses Merkle trees for anti-entropy. If the ring rebalances mid-repair, the token ranges have shifted, so the comparison has to handle the new ownership... I'd guess Cassandra either pauses repair during rebalance, or it re-computes the Merkle tree on the new owner. I'd verify in the docs." — This is the senior-engineer answer. Reasoning from first principles + admission of uncertainty + plan for resolution.

Bluffing is the single most common reason candidates fail at staff+ levels. Senior interviewers can spot it instantly. They respect "I don't know but here's how I'd find out."

Interviewers at most FAANG companies use a rubric like this. Candidates who know it perform measurably better.

1. Skipping requirements. Diving straight into "I'll use a load balancer..." Interviewer immediately downgrades.
2. Over-engineering for hypothetical scale. "I'll use Cassandra and Kafka" for a problem that fits in one Postgres. Show you can match design to actual scale.
3. Going deep on everything. Burning 30 minutes on the API design; never reaching the deep dive. Pick ONE rabbit hole.
4. No numbers. "We'll cache it" — how big a cache? "We'll shard it" — how many shards? The numbers prove you're not hand-waving.
5. Defensive when challenged. Interviewer says "what if X?" — they're testing flexibility, not attacking. Engage: "Good point, I hadn't considered that. Let me think..."
6. Ignoring interviewer hints. They keep asking about consistency. They want you to discuss CAP. Take the hint.
7. No production thinking. Beautiful diagram; no thought to deploys, monitoring, oncall. "How would you debug this at 3am?" should always have an answer.
8. Time mismanagement. 50 minutes in, just starting the deep dive. Game over.

Use the framework on every problem in this portfolio. The 11 sections of each problem page mirror the 60-minute interview shape. Practice driving through them in order, with time discipline. Every concept page is the depth-on-one-thing you'd reach for in the deep dive.