Design Top-K System — Section 5: Interviewer Q&A (Shallow Deep Dive)
Goal of this section: Simulate common system design interview questions and demonstrate strong, concise Staff/Principal-level answers.
Q1. Why not compute Top-K on every request?
Answer:
Computing Top-K on read requires scanning and sorting a large candidate set, which is expensive and unpredictable at scale. Since Top-K queries are:
- Repetitive
- Highly skewed
- Read-heavy
We precompute and materialize Top-K lists per scope and window. This shifts cost to the write path and enables:
- Stable latency
- Aggressive caching
- Simpler serving logic
Sign in to continue reading
The rest of this lesson is available with a free account. Signing in with Google or Microsoft is free.
Sign in to read the full lesson