System Design: Leaderboard Service (A) and Real-Time Messenger (B)

A) Leaderboard Service for a Large-Scale Game

Design a service to:

1. Compute a user’s global rank and their rank within their friendship circle.
2. Retrieve the k users immediately before and after a given user, both globally and within the friend circle ("around me").
3. Handle high write rates, pagination windows, anti-cheat, seasonal resets, and multiple game modes.
4. Detail:
   • APIs (read/write, pagination, friend-scope queries).
   • Data models (e.g., sorted sets, secondary indexes) and schema.
   • Sharding/partitioning strategy.
   • Caching strategy.
   • Consistency model and rank computation approach.
   • Friend-graph integration at read time and/or via materialization.
   • Backfill/batch workflows (rebuilds, seasonal archives).

Assume tens of millions of active players per season and high write rates (hundreds of thousands of updates per second) with low-latency reads.

B) Real-Time Messenger Application

Design a system to support:

1. 1:1 and group chat with correct per-conversation message ordering.
2. Delivery guarantees: choose among at-least-once, at-most-once, or exactly-once; justify the choice.
3. Online presence, typing indicators, read receipts.
4. Media sharing (images/video), push notifications, end-to-end encryption (E2EE), offline sync, and fanout for large groups.
5. Spam/abuse controls and disaster recovery.
6. Provide an end-to-end architecture, storage choices, protocols, scale estimates, and operational considerations.

Assume hundreds of millions of DAU, multi-region deployment, and peak traffic in the hundreds of thousands of messages per second.