The problems with data engineering you cite are very real, but I’m afraid the antidote you suggest may be worse than the disease. Having worked without any data engineering at all, and also with data engineering that acted as a bureaucratic data gate-keeping department, there is a middle ground. Done right data engineering can make the job of data science faster, more efficient and error-free. The trouble is that in most organizations the data organization and infrastructure was designed for the needs a reporting function that is rigid and inflexible, not for data science that has completely different needs. It’s not the position/discipline that is the root of the problem, it’s bureaucracy and hubris.