Defining the quality of a machine learning production system

Building a machine learning system and maintaining it in production are two very different things. Some folks over at Google wrote a paper that shares their thoughts around all the items you might want to test or check for your production ML system. Relevant links:

Popout Listen on the new Podbay