DataStaR, a Data Staging Repository (http://datastar.mannlib.cornell.edu/) in development at Cornell University’s Albert R. Mann Library, is intended to support collaboration and data sharing among researchers during the research process, and to promote publishing or archiving data and high-quality metadata to discipline-specific data centers and/or institutional repositories. Researchers may store and share data with selected colleagues, select a repository for data publication, create high quality metadata in the formats required by external repositories and Cornell’s institutional repository, and obtain help from data librarians with any of these tasks. To facilitate cross-domain interoperability and flexibility in metadata management, we employ semantic web technologies as part of DataStaR’s metadata infrastructure. In this paper, we describe the overall design of the system, our work to date with Cornell researchers and their data sets, and possibilities for extending DataStaR for use in international agriculture research.