OK, I did something cool today. I took some basic/naive #Kafka stream processing code in .Net (consume-process-produce), and through microbatching these with transactions massively increased throughput, from 500msg/s to 25,000msg/s with a _single thread_. No librdkafka tuning. Programming is about understanding your primitives, in this case network interactions.

I rarely do technical work any more, focusing on tech strategy, so it's fantastic to get that sweet dopamine hit from problem solving.