Protobuf and Avro decode ~20x faster than JSON in a microbenchmark and ~7x faster in a real pipeline. The popular explanation — "no keys in the payload" — is only half the story.

Most devs barely touch their personal laptop. Mine is now a tiny homelab I never look at — kind cluster today, maybe Photos and Ollama later.

Added tracing to the sink component and experimented with Go profiling to understand performance bottlenecks

Experiments with Kafka and Go to check performance and required configurations