Interesting article indeed. And yes, it is great to see Auto ML getting some traction in the DS community as it cuts down the not so exciting part of a DS role (heavy scripting for data prep’ and feature engineering, hyper-parameters tuning) to then get more into the really interesting part (surface meaningful and reliable predictive insights, iterate with business stakeholders to address the pain and deliver a model that devops integrate seamlessly). There is one part that, I feel, is not explicitly mentioned in this good post which is the need to leverage the cloud. This is one of the reason PredicSis.ai got ported onto AWS marketplace (disclaimer: I am part of the team). The usual pains related to selecting, installing, validating the right auto ML solution are addressed by using a solution that can simply be activated in a couple of clicks on the AWS marketplace. Another technical hurdle that is removed and, again, helps DS focus on what they are good at.