[wackspurt]

>>> We apply cutting edge machine learning techniques on petabytes of data and thousands of Spark/YARN nodes.

I'm curious. Do you use an open-source Java ML library (dl4j, etc.) or do you have an in-house implementation?