hello. I am an ML Engineer who is actively using bentoml for model serving. I am wondering if there is a best practice for configuring serving based on bentoml to provide API while adding the ability to consume Kafka messages. Two types of inference requests are required: http/grpc API request and Kafka message. Although it would be ideal to have a separate server that consumes messages in front of bentoml, the internally configured CD is designed to only serve bentoml, so it would be nice to be able to add kafka consume to the bentoml service. Thank you in advance for your reply. 🙂

in our team we setup a kafka consumer inside a service.on_startup hook