Patterns

  • Geo-Proximity Pattern
  • Recommendation Pattern
  • Pub-Sub Fanout Pattern

Expected topics

  • Tinder
  • Geo-Proximity Pattern
  • Recommendation Pattern
  • Pub-Sub Fanout Pattern
  • M+ users
  • pairwise
  • pool
  • A like B
  • B like A
  • match

Self-check prompts

  • What users, scale, latency, availability, and consistency requirements should you clarify for Tinder?
  • What are the main APIs, data model, and request flow?
  • Where is the main bottleneck around Geo-Proximity Pattern, Recommendation Pattern, Pub-Sub Fanout Pattern, and how would you scale it?
  • What failure mode matters most, and how do retry, recovery, and idempotency work?
  • Which trade-off would you choose, what do you lose, and when would you change that decision?

Common mistakes

  • Jumping into vendor names before clarifying requirements and scale.
  • Listing components without explaining the end-to-end request flow.
  • Leaving the bottleneck vague instead of quantifying capacity, partitioning, and recovery behavior.
  • Mentioning trade-offs without choosing an option and explaining the condition that would change the decision.