Based on what I read it's very similar to Kafka Streams + batteries ([semi]automatic workload orchestration, reactive queries, higher-level/slicker/"smarter" AP

Based on what I read it's very similar to Kafka Streams + batteries ([semi]automatic workload orchestration, reactive queries, higher-level/slicker/"smarter" API (?)) Could you please compare Rama with Kafka Streams, especially from the point of view, if I would try to reimplement Rama API on top of Kafka Streams? What fundamental difficulties I'd face?