Question

Design schema governance for an event platform with ~600 Avro/Protobuf event types flowing through Kafka, where producers and consumers are owned by ~80 independent teams that deploy on their own cadence. Today, a producer adding a required field or changing a type silently breaks consumers, causing deserialization storms and dropped events discovered hours later. Design a system that lets producers evolve schemas safely without lock-step coordination, while guaranteeing existing consumers keep working.