Benchmarks

• Fast is the easy part. Here is why you can trust KPipe with your offsets.
• It is 2026. Why are your object mappings still strings?
• Benchmarking KPipe against the parallel-Kafka libraries you would actually pick