I'd argue that any ML project of that scale is a monumental task, and without good organization and resources, people are bound to burn out. Unfortunately with

I'd argue that any ML project of that scale is a monumental task, and without good organization and resources, people are bound to burn out. Unfortunately with ML, there's only so much one can do with just code. The many loops of hypothesis testing that involve endless training hours are just exhausting since the response is just so far away and involves constant context switching.