Project ProtoStar is a new initiative bringing together universities, research organizations, industry, and the open science community to make publicly funded scientific research easier to discover, reuse, and build upon. The project specifically targets the release of datasets, models, and code in AI-ready, machine-actionable formats to unlock their potential for artificial intelligence-driven research. Founding partners include Carnegie Mellon University, Purdue University, Georgia Institute of Technology, Arizona State University, and The George Washington University, operating through the Open Forum for AI (OFAI). Additional key partners include the Astera Institute, Hugging Face as the infrastructure provider, and the Open Research Community Accelerator (ORCA) via its Fostering Innovation Network.
The initiative addresses a critical gap in current scientific sharing practices. While the U.S. federal government invests approximately $190 billion annually in research and development with increasing expectations for public access, simply making data available does not ensure usability. Scientific artifacts are often scattered across incompatible systems, stored in legacy formats, or released without necessary documentation and metadata. These limitations hinder researchers from building on existing work and prevent AI systems from effectively discovering and combining resources across disciplines. Project ProtoStar aims to close this gap by encouraging systematic sharing supported by open infrastructure and practical publishing guidance.
The demand for reusable scientific resources is already evident on platforms like Hugging Face, which hosts over 17,000 public datasets in core scientific domains such as genomics, chemistry, and climate science. This number approaches 29,000 when including medicine and mathematics. The pace of sharing is accelerating, with researchers uploading more than three times as many new science datasets in the first nine months of 2026 compared to the same period in 2025. Documentation plays a crucial role in this ecosystem, as 90% of dataset downloads on Hugging Face come from fully documented resources. Existing repositories from organizations like the NIH's National Center for Biotechnology Information and NASA IMPACT have already generated millions of downloads.
To demonstrate the value of these AI-ready resources, Project ProtoStar will organize a series of open challenges hosted by participating institutions. The inaugural challenge, titled "Climate Risk Intelligence," is hosted by Carnegie Mellon University and runs from November 2 to November 13, 2026. Participants are invited to test whether AI can improve predictions of urban climate conditions using historical weather data from 210 U.S. cities. Teams will use records through 2018 to predict actual conditions from 2019 to 2026, forecasting indicators such as average annual temperature, total precipitation, the hottest day of the year, and cooling degree days.
Beyond technical forecasting, the challenge requires participants to propose a short climate adaptation policy idea informed by their data analysis. This component encourages connecting scientific findings with practical public benefits, such as preparing cities for extreme heat or managing changing precipitation patterns. Teams will submit their outputs as Hugging Face datasets, with performance assessed via a public leaderboard comparing prediction errors against statistical baselines. Final forecasting results are scheduled for publication in early 2027 once complete weather records for 2026 are available.
Participating organizations formalize their involvement through a memorandum of understanding to release suitable scientific artifacts in AI-ready formats on open infrastructure. The initiative seeks to generate practical evidence about the infrastructure and practices needed to support open science at scale. Researchers, developers, students, and companies are invited to participate in the challenges, while universities, federal agencies, and national laboratories interested in joining Project ProtoStar can contact the organizers at protostar@orcaopen.org.
Sources: Hugging Face Blog.