We faced exactly this problem at the current company I work for. When the company was just starting, we had the great idea to use interview questions "like the

We faced exactly this problem at the current company I work for. When the company was just starting, we had the great idea to use interview questions "like the ones they make in SV", thus used algorithmics-heavy problems when interviewing candidates. Fast forward 4 years, and I realized that the dev team we have is basically a group of just-graduated Engineers or people with at most 1 year of experience. My "eye opener" moment was when I realized that by asking for difficult algorithms (tree and graph traversal, subsequences, etc) the following was happening: The people that know the most about how to solve these types of problems are students, or people that had graduated recently. Not only that, but this groups is a very reduced sub-group from the recent graduates group, because these people dedicate their Uni free time to solve these problems and get in ACM, IOI competitions. Thus, they don't play with different technologies, they oftentimes don't feel confortable with shell, and finally, the type of systems they develop are optimized to be "used and thrown away".