Hugging Face, five US universities and partners announced Project ProtoStar, an initiative to release research data, models and code in AI-ready formats, starting with a Carnegie Mellon climate challenge on November 2.
Hugging Face announced Project ProtoStar on October 29, an initiative meant to make publicly funded scientific datasets, models and code easier to discover, reuse and build upon, particularly for AI-driven research. The founding partners named in the announcement are Carnegie Mellon University, Purdue University, Georgia Institute of Technology, Arizona State University and The George Washington University, working through the Open Forum for AI; the Astera Institute; Hugging Face as infrastructure partner; and the Open Research Community Accelerator as community partner. Participating organizations sign a memorandum of understanding supporting the release of suitable research artifacts in AI-ready, machine-actionable formats on open infrastructure.
Hugging Face frames the project as a response to a usability gap: the US federal government invests roughly $190 billion a year in research and development, and agencies such as the NIH and NSF increasingly expect results and data to be shared, yet files are often scattered across incompatible systems or lack documentation and provenance. The company says it hosts more than 17,000 public datasets across core scientific domains, nearly 29,000 when medicine and mathematics are included, and that researchers shared more than three times as many new science datasets in the first nine months of 2026 as in the same period of 2025. It adds that NCBI's 18 repositories on the platform have drawn more than 11 million downloads and NASA IMPACT's 27 repositories about 770,000, as of October 26. These figures come from Hugging Face's own hub data.
The first of a series of open challenges, hosted by Carnegie Mellon University, runs from November 2 to 13, 2026. Participants will use historical weather records for 210 US cities through 2018 to predict actual conditions from 2019 to 2026 on four indicators: average annual temperature, total annual precipitation, the hottest day of the year and cooling degree days. Forecasts will be scored on a public leaderboard against simple baselines such as historical averages and linear trends, and teams will also propose a short climate adaptation policy idea. Final forecasting results are due in early 2027, once complete 2026 weather records are available. Organizations that want to join can contact protostar@orcaopen.org, according to the announcement.
