Notes from building backend and AI infrastructure — high-throughput systems, narrated in my own voice.https://mukundtandon.dev/https://mukundtandon.dev/blog/six-hundred-thousand-a-second/https://mukundtandon.dev/blog/six-hundred-thousand-a-second/Building a Go log ingestor that scales from ~16K to over a million logs per second — a three-stage Kafka → buffer → ClickHouse pipeline, and the config changes that unlocked the throughput.Sun, 15 Mar 2026 00:00:00 GMTdistributed systemsgoclickhousekafka