How the interviewer opens it
“Design a question-and-answer platform like Quora. Users ask questions, write answers, vote, and follow topics or people. Help users find relevant questions and useful answers.”
The prompt is vague on purpose. The first minutes are for you. Reduce the prompt to the core features and agree on the numbers before you draw.
What to clarify first
- Users and existing content
- Read traffic
- Write traffic and content size
- Content and follows
- Vote rules
- Discovery
- Latency
- Consistency and availability
- Durability and regions
- Scope limits
The interviewer has the answers. In a SysRound round, Alex gives these answers, as a real interviewer does.
Scope of a strong answer
Core requirements
- Publish questions and answers with topic labels.
- Vote on answers and read answers in rank order.
- Follow authors, topics, and questions.
- Find relevant questions through a personal feed and text search.
Usually out of scope
- Ads and payments.
- Comments and private groups.
- Model training.
- The moderation review process.
- Full abuse detection.
What interviewers ask about
- Relevant question discovery
- Users follow several kinds of interests. Popular topics create many updates, while repeated answers can make the same question dominate a feed.
- Vote correctness and answer quality
- Vote changes must remain correct after retries. Answer order must also give useful new answers a chance without favoring raw popularity alone.
- Freshness and visibility
- A question or answer appears in several public views. Edits and removals must reach those views despite delays and failed components.
Common mistakes
- Treats discovery as a full database scan or an unexplained recommendation model.
- Assumes every topic update should immediately reach every follower.
- Stores vote totals without the user's durable vote state.
- Equates answer quality with raw vote count.
- Uses one consistency rule for content, feeds, vote state, and aggregate totals.
- Removes content from the primary store but ignores search results and cached previews.
The bar by level
- Senior (L5/E5)
- Controls candidate retrieval and topic fanout cost. Handles vote retries, stable feed pages, event recovery, and content removal across derived views.
- Staff (L6+/E6+)
- Quantifies fanout and popular-answer limits. Defines correctness boundaries during failures. Explains regional ownership, derived-view rebuilds, and quality safeguards as requirements change.