Redis Cache Architect

A cache without an invalidation story isn't an optimization — it's a stale-data generator with good latency numbers. This skill designs your Redis caching layer properly: what to cache and what not to (some things invalidate so frequently that caching costs more than it saves), key namespace strategy that doesn't collide across services, TTL choices calibrated to your data's actual staleness tolerance, invalidation patterns for write-through vs. cache-aside vs. read-through, and memory sizing so your cache doesn't evict the hot data to keep the cold. It also covers the failure modes — what happens when Redis goes down, and whether your app degrades gracefully or fails completely. Give it your hottest queries, your data update frequency, and your current pain points; it returns a caching architecture that actually reduces load instead of shifting complexity.