Just put into github a stackoverflow client for Java. Access to stackoverflow data is via Java objects that hide server calls to stackoverflow and which cache retrieved data in a local Elasticsearch index. Calls to the server are further reduced by automatic pagination and a governor that limits calls to 25 per sec.
I have more stackoverflow objects to add, but would love to get feedback on what's there.
The code is structured as a mavenized Eclipse project and is currently used by my yet-to-be-released Storm topology that builds and applies models to predict which newly asked questions a particular user should answer.