Fetching a user's feed—complete with tags, media, author details, and like counts—is an expensive relational query. To achieve our goal of sub-100ms API response times, we cannot hit the SQL database for every page load. We have implemented a distributed, in-memory caching layer using Redis. When a feed is requested, the system first checks the Redis cache. If a cache miss occurs, it queries the database, formats the JSON payload, and stores it in Redis with a Time-To-Live (TTL) expiration. We employ an event-driven cache invalidation strategy so that when a new post is created, the specific cache keys are flushed, guaranteeing users see fresh content instantly.

Fetching a user's feed—complete with tags, media, author details, and like counts—is an expensive relational query. To achieve our goal of sub-100ms API response times, we cannot hit the SQL database for every page load. We have implemented a distributed, in-memory caching layer using Redis. When a feed is requested, the system first checks the Redis cache. If a cache miss occurs, it queries the database, formats the JSON payload, and stores it in Redis with a Time-To-Live (TTL) expiration. We employ an event-driven cache invalidation strategy so that when a new post is created, the specific cache keys are flushed, guaranteeing users see fresh content instantly.