My research was on spatial query processing -- z-order-based techniques. The idea is to map spatial objects to one dimension by noting how a space-filling curve (z-order) passes in and out of the object. Overlapping objects can then be found by simple operations on 1-d data structures supporting random and sequential access, (e.g. sorted array, btree). I believe that the same transformation is used in geohashes.

The product is Akiban Server, a database system. I implemented nearest neighbor and point containment queries, and integrated it with our query optimizer. (You can download it from akiban.com and try it out.)