Netflix Onsite Interview: Full Loop Experience (Coding, System Design, Product Sense)

Question

**Format:** Full onsite loop — typically 5-6 rounds covering coding, system design, behavioral, and product sense.

**Coding Round**
- Data stream problem: given a stream of watch events, compute the running median watch duration per genre.
- Follow-up: handle late-arriving events within a 5-minute grace window.

**System Design Round**
- Design the Netflix homepage recommendation service.
- Discuss: content catalog storage, real-time vs. batch recommendations, CDN edge caching for thumbnails, fallback when ML model is unavailable.

**Product Sense Round**
- Netflix adds a "Social Viewing" feature (watch together remotely). How do you measure success?
- Clarify: leading vs. lagging metrics; how to detect if the feature is cannibalizing solo viewing sessions.

**Behavioral Round**
- Describe a time you disagreed with a product decision and how you handled it.
- Tell me about a system you built that failed in production — what did you learn?

## Follow-ups

1. In the streaming median problem, what is the time complexity of each insert if you use two heaps?
2. For the homepage service design, where would you place the personalization layer relative to the CDN?
3. How do you distinguish a metric drop caused by the new feature vs. an external event (e.g., a major sporting event)?

Netflix Onsite Interview: Full Loop Experience (Coding, System Design, Product Sense)

Interview Experience

Full Details

Topics

More from Netflix