So for me portfolios are for junior devs & graduate students looking to leave academia. Therefore, I look for things that are uncommon in those environments, engineering over theory, documentation & tests over novelty.
Note: if I were hiring for a researcher the above would not be true.
What I do have is a bunch of great reviews from my managers in the past years. Do those carry any weight in an interview?
- If I had a good work sample based proxy for hiring, I'd use that and basically nothing else.
- But that never happens, so what I (and most places I've encountered) actually use is network based hiring. So a bunch of great reviews from your managers is awesome for hiring assuming I know them. If I don't its basically a crap shoot, but I view it as much more valuable than some half baked collection of github repos proving to me you've done the the first couple chapters of several hot "how to" books.
In a should be case you'd have a hiring pipeline that had a good objective work sample proxy for the work you were trying to do and you'd judge blind other than that.
In the real world, I've found what actually happens is network based hiring. I hire the people I already know are good. After that, its mostly a crap shoot but I'd take popular open source contributors over other candidates all things being equal (though by the time you've contributed to a popular open source project you are dramatically past the bar for hiring outside of the "network" tract).