I asked about replacing the Thrift collector payload a while ago while discussing the Clojure collector replacement: Replacing the Clojure Collector
If anything, I would expect Snowplow to leverage Iglu Central and JSON Schemas instead of Confluent’s Schema Registry and Avro. As far as I know JSON is fully supported by Kafka Connect too (see key.converter in https://kafka.apache.org/documentation/#connect_running).