Writing
Notes on Redis, real-time data, and AI systems.
Cut your LLM bill with semantic caching (video)
A live walkthrough of Redis LangCache: serve semantically repeated questions from cache in milliseconds, with the token savings adding up on screen.
Why your online feature store belongs in Redis, not DynamoDB
Fan-out and the p99 tail are what break online feature serving, and why Redis sits above the warehouse as the hot tier.
Rate limiting LLMs with a token bucket in Redis
A gateway and one Lua script turn a runaway token bill into a control plane.