It's interesting how it gets Jupyter working from streaming workloads. Jupyter notebooks are the de-facto choice for most data scientists around me. Haven't really tried it out myself but I imagine this could be a value addition for streaming ML use cases for sure.