Actually, serving ML inferences was the initial motivation for building the library. We couldn’t find anything similar to FastAPI for Apache Kafka and then we decided to build one.