The article make notable mistakes:<p>1) Yes, you have more/big data, but what you need is labeled data which is different. I.e. it does not matter if I have 100 picture of cats or 100000 picture of cats, if this is the same cat.<p>2) Feature store are not for feature engineering, more for feature discovery and sharing, feature engineering is a different area of concern.<p>3) Not sure about the fact that data changes so much. Yes some domain it might , but some it does not.