Hubski has no database. Instead, data is stored in flat files as s-expressions (lists, it's Lisp!). All the data you see here is actually in memory, loaded on startup, plus newly added data. Tags have existed as elements within post and user data, but I am currently creating an independent directory for them, so they can have their own associated elements that can be updated, sorted more quickly, etc.
There's definitely some work to be done with all this as things progress. It should be fun.