Hey Phil, I’ve worked with the HTM engine (https://github.com/numenta/numenta-apps/tree/master/htmengine) which sets up several services that listen to data streams, HTM models are swapped into memory from disk when new data comes in (and out when the data stops). While there’s not a service explicitly written for Kafka processing that I know of, it wouldn’t been to hard to write one. For example, I’ve taken Austin’s skeleton HTM engine app, and fleshed out this simple web service (https://github.com/htm-community/skeleton-htmengine-app/blob/master/webapp.py) for a production app before. So each time you consume from Kafka, there would be some calls to the “API” that HTM engine provides. (Here’s a short video where I talk about some work I did with HTM engine https://youtu.be/iJZSAJmg4OE?t=1h56m23s). Would be happy to discuss further.