As a non-computer science person, I feel very dumb right now. :-(
edit: cause I don't remember much of this stuff.
You use it so much that it is down to intuition, or you never use it at all.
For example, my relationship to Orders of Complexity has been reduced to sorting options in my head, and picking the first one that is simple enough to keep working, and fast enough to satisfy the current and estimated future scope of the problem (which is also intuitive Capacity Planning).
If you make me explain the decision it might sound like I picked one at random.
Are you asking if I have written (shipped) a sort implementation? Only once, ages ago. No built-in stable sort, and none of the dozen articles I found had examples that scaled past 20 elements (which I found quite upsetting). I did my own port of mergesort that was almost 2 orders of magnitude faster. Constant complexity kills.
It's good to build a few on your own to become an informed consumer. Just don't 1) ship it or 2) use it as an interview question. If you want NIH people, you get them by asking NIH questions during the interview process.