The other thing here is, of course, that these "companies we all know" are not the be all and end all for software development. There are hundreds of thousands

The other thing here is, of course, that these "companies we all know" are not the be all and end all for software development. There are hundreds of thousands of smaller companies the world over doing development in every field you could imagine. I know nothing about them, but some of them are very likely to be doing things I would find fascinating. There's very little to connect me with them except by chance. Even if I'm aware of the main companies in the specific field I work in, there are still far more that I'm unaware of. That's a lot of missed opportunities. The current process is inefficient and suboptimal for all concerned. Recruiters have trouble identifying good candidates. And on the other side we have trouble identifying good companies/projects. On both sides we dumb down the skills provided/needed to bullet points of technologies and tools, often completely ignoring the actual problem domains. While I don't think the suggestion here about gleaning insights from git repositories is a solution by any means, it might make a small improvement by providing some deeper insight into real demonstrated skills and experience. As someone else mentioned, of course this could lead to gaming the system, but I suspect most fraudulent stuff could be filtered out--there would be clear differences between real projects contributed to over the course of years and "resumé fodder".