Hello Lee,

Thanks for understanding my perspective. The way I see it is that as the field matures, companies will start forming teams rather than ask for a 'full stack data scientist' . As you rightly said, 'full stack data scientist ' is just a band aid but this band aid deepens the wound.

Also, automating model building is another big problem in the field currently. Many are just fixated on the final accuracy or F1 scores but forget that the algorithm makes no sense to apply in that domain.

All the data science library salespeople will say 'data science is merely a data engineering / software problem. But the truth is far from it. In my opinion, feature selection and modeling can't be automated. It takes domain expertise do it successfully.

This automating the modeling part again comes from companies who feel they are paying their data scientist 'too much'. Hence they want to do away with them or at least mitigate their importance. Little do they understand that, they are going after a wrong problem.

Data Science once done correctly will only require minor maintenance over time not a complete overhaul. But Data Engineering on the other side often breaks due to bugs, technical debt etc. So Data engineering is kind of evergreen, there are always problems and hence always 'work to do'. This is also another reason why many are now pivoting to data engineer roles rather than data science roles.