- Built a matchmaking engine on Redis Streams and BullMQ consumer groups, enabling HTTP, Socket, and Worker
nodes to scale horizontally with no shared state
- Designed idempotent jobs with safe retries and consumer rebalancing, ensuring crashed workers recover without
duplicating matches or corrupting queue state
- Architected a two-phase scaling strategy using Redis Streams for early growth and Kafka with partitioned topics for
a projected 10M+ user global deploymen