There is probably a need to bridge the gap between machine learning and software engineering. Building a standalone model is great, but how about also considering the engineering challenges when it needs to be deployed as a part of a system into production?
Your thoughts?