I'll state the obvious and answer with No. There are not enough skilled software architects to go around and many who consider themselves skilled are not actual

I'll state the obvious and answer with No. There are not enough skilled software architects to go around and many who consider themselves skilled are not actually producing good code themselves, probably including many confident posters here in this forum. The idiosyncrasies and tastes of many 'senior' software engineers would likely make the code unreadable and unmaintainable for the average scientist and possibly discourage them from programming altogether. Software architecture is an unsolved problem as evident in the frequent fundamental discussions about even trivial things, highlighted by a cambrian explosion of frameworks who try to help herding cats, and made obvious in senior programmers struggling to get a handle on moderately complex code. I propose scientists keep their code base as simple as possible, review the code along with the ideas with their peers, maybe use Jupyter notebooks to show the iterations and keep intermediate steps, and, as others state, show the code as appropriate and try to keep it running. There is no silver bullet and very few programmers could walk into your lab or office and really clean things up the way you'd hope.