It is great to have more examples of building a indexed fast query representations of the the current ledger state.

What did you guys go with as a policy for handling or avoiding orphan blocks...

Currently we're saving every block we see. And if we find an orphan's parent which is on the main chain we try to connect them both. We aren't doing anything to explicitly avoid them.

If you're referring to side chain blocks (which some people refer to as orphan blocks) we'll extend them to be the main chain should they win or leave them forever marked as side chain blocks. We aren't doing anything to explicitly avoid these either.

Why Ruby? (Nothing against it, just curious)

We're mostly a ruby shop here - so using what we know is more efficient and allows more engineers to contribute (both internally, and from an open source perspective).