> No postdocs should still be learning basic skills. The (ostensible) purpose of a postdoc is a final step to prepare very talented PhDs for a faculty position. People who are faculty material should have mastered all relevant skills during their PhD.
That attitude is very bureaucratic.
Computational research is always interdisciplinary. Administrators organize universities into departments according to the needs of undergraduate curricula, but those administrative boundaries have little to do with the skills needed in research. The skills you learn during PhD are probably enough for your PhD research, but then you may need new skills when you start a new project as a postdoc. And again when you start a new project as a junior faculty member.
Postdoc is also the career stage where many people switch fields. For example, a background in X or even computational X is often insufficient for some research topics in computational X. You need to hire people who got their PhDs in CS, mathematics, or statistics, and then those people must learn X before they can contribute. And once they start contributing, they may have to learn software engineering to turn the methods they have developed into useful tools.
That attitude is very bureaucratic.
Computational research is always interdisciplinary. Administrators organize universities into departments according to the needs of undergraduate curricula, but those administrative boundaries have little to do with the skills needed in research. The skills you learn during PhD are probably enough for your PhD research, but then you may need new skills when you start a new project as a postdoc. And again when you start a new project as a junior faculty member.
Postdoc is also the career stage where many people switch fields. For example, a background in X or even computational X is often insufficient for some research topics in computational X. You need to hire people who got their PhDs in CS, mathematics, or statistics, and then those people must learn X before they can contribute. And once they start contributing, they may have to learn software engineering to turn the methods they have developed into useful tools.