Wrote about how I built a text-to-speech system for 25K daily users without melting a server. Lots of lessons around caching, concurrency, and audio processing. link in the comments. https://t.co/l3hzjBBOZB
