Most open source projects are pretty bad at offering structure in general. The big exception to this is when projects participate in the Google Summer of Code, where Google mandates a set of deadlines and metrics to be used to measure success, and success determines how much Google pays out to the host organization and to the student. (And in turn mentors tend to mostly self-select for being able to provide that kind of structure.)

Do you want to find a project where you're doing standalone SQL work, or is there a particular programming environment that you would like to use SQL with?

As far as structure and having clear user demands, I've always found doing projects for friends pretty helpful. I might never finish, but I'll probably get at least something done.

If you can find a research group or something like to work with, that might be even better, as large numbers of scientists know how to gather lots of data and maybe even look at specific properties of data, but know little about information management and working with databases, and so it's easy to have a big impact there. I wish that corivax's old lab still had funding, since if those projects were still ongoing, I'd have some pretty concrete suggestions of easy SQL things you could help with that would contribute to fairly groundbreaking research.