I somewhat disagree here, but also somewhat agree.

In my org, we get a lot of requirements that require very different skillsets. For the first 2-3 years, our task list was mostly CRUD stuff with some domain specific logic, but otherwise a boring web app. In the last 1-2 years, we have:

• ported a Fortran simulation to Python
• embedded a C++ simulation in Python
• created a 3D UX for our previously 2D only app (lots of 3D logic on both FE and BE)
• implemented a machine learning algorithm to train our simulations

If I hired only for the work I’d seen in the past, we’d be completely unfit to handle this workload since we’d mostly have people who are really good at building CRUD apps (so DB optimization and quick UX building).

On the flipside, we cut off huge swaths of work so people don’t need to wear too many hats. We have:

• dedicated devOPs - handles everything from trst pipelines to prod deployments
• dedicated QA - manual and automated app-level testing - devs still do unit testing
• dedicated product teams who handle feature requirements and documentation
• dedicated UX team to produce designs for FE engineers to implement

So our devs only need to worry about development, but they also need a broad skillset in that domain, from everything from local tooling to working in different domains. We hire a diverse set of candidates, some with a heavy math background, some with design experience, and some with low level programming experience, because we never know what projects we’ll get or who will suddenly leave the org.