The vast majority of my web development experience is in PHP; this has far less to do with it being a particularly elegant language, and more to do with the fact that it is widely supported and super-simple to use as the basis for hobby-site development. I've thrown together a few sites in Pylons and Ruby on Rails — none of which are featured here, due to the fact that their content quickly became irrelevant — and have seen Tomcat once or twice. While I used to perpetually bash Javascript, I've since (half a decade ago or more) recanted my objections and really enjoy the modern flavor of Javascript used to make sites more, not less, usable.
To get started, here's a list of more-or-less active web projects. Nothing here is at risk of being bought by Google, but they certainly keep me amused.
Sometimes a picture is so awesome, it deserves its own domain. It's useful to have a throwaway site to mess with SEO and the like. The good news is, we're now ranked third in Google for "acid monkey" (even without the quotation marks), and ranked second for "monkey on acid". Count this one a victory.
One of the greatest things about working in the Bay Area is the sheer amount of quality Mexican food. On the other hand, it's also a curse — you're never quite sure exactly which hole-in-the-wall taqueria you should go to for optimal grub. Enter Burratr; description will not do it justice. This is a community project, for people who are trying to learn the whole PHP + AJAX kind of thing, so be forgiving; and for once, I'm some kind of mentor.
My extended family is large enough to need its own domain. Being the computer guy in the group, I was drafted to design some sort of photo-sharing site. While this didn't work out in the long run, it has provided some interesting insight into database design and the like — this was the first project I'd ever embarked upon after learning from some of the DBAs at work. There's a cool standalone Java app to upload pictures, since the servers (1&1) are too slow to permit large file uploads via PHP, Perl, or whatever. It's a reasonable site, but sometimes technology loses to personal politics.
A couple of friends wanted to see if we could put together a good working group, to develop a site which would sport a dandy little application and attract a decent user base without too much overhead. After a little brainstorming, we hit on the idea of a site which would allow the recently-engaged to show off their new rocks; perhaps this was a function of one or two in the coalition having popped the question. The group has since largely disintegrated, but the site remains in decent working order.
At one point the proud owner of the only student-run bar on campus, it seemed fitting to auction off the related bar-aphenalia in a truly outstanding fashion. Call this one an experiment with horizontal design, and a lesson as to why it just doesn't work all that well. AJAX and picture scrolling at its finest, if I do say so myself. -2 points for bad layout, though.
The site was awesome and won some national award or another when it was under my control (Best Orientation Website, as awarded by the National Orientation Directors Association). Since then, the site itself has devolved into ruin, so it's best for all of us if I just link to the archive.org cache from back in the day. Awesome fact: we couldn't budget for the university font at the time, so I made the "W" all by my lonesome.
If any, this is the site which is closest to my heart. Being big into cycling, and big into not hanging around the same part of the country for too long, I've often found myself at a loss for good biking routes. Routeslipper solves that problem fairly elegantly, but unfortunately is just a little bit too Web2.0 to succeed (asking users to manually mark relevant cycling intersections and trace roads via Google Maps). I like to think it stands a fighting chance, since it's definitely a nifty little tool, but the layout is back from my days of noncompliance (we've all been there), and quite frankly there are other sites which make it much easier to upload your routes, even if they provide less utility to the end-user.
In these modern times, every official-couple-to-be needs a site to offer guests information and — perhaps more importantly — information on the innumerable but very specific things they may buy you to make you happy. It's not a particularly fancy site (although there's some neat backend stuff monitoring the gift registry status, check out the graph) but it's reasonably pretty and acts as a good website should: offering information without overload or too much cruft.
Occasionally — but not too often — I embark on actual software projects. Although I'm much more comfortable with a web-based development cycle (especially rapid deployment and bug fixing), actual client applications are still better at quite a lot of things that I am (academically) interested in.
GemIdent is not my idea, but I put many loving nights into its codebase. The purpose of GemIdent is cancer cell identification, and it classifies fairly well. Although the project has since been open-sourced, its root was funded by the Stanford Medical School.