Highlights from GlobusWorld 2025

GlobusWorld 2025, now in its 15th year, brought together over 130 researchers, system administrators, developers, and IT leaders from 42 top research computing centers, national labs, federal agencies and universities from around the world. The conference is a forum for insights into new methods of managing research storage and computation at scale, and building cutting-edge data portals and science gateways to advance scientific research.

After a morning of technical tutorials, the conference kicked off with a welcome address by Globus co-founder Ian Foster, followed by an opening keynote by Rachana Ananthakrishnan, executive director and head of products at Globus. Ananthakrishnan described how Globus, with reliable and secure hosted, persistent, scalable and resilient services, combined with local agents and plug-ins, delivers the services required to overcome many data management hurdles researchers face in today’s globally distributed IT environments. She provided many real-world examples that demonstrate how Globus is used within the US national laboratories and leading research organizations to accelerate workflows and pipelines for data analysis. She described new features and future product plans, including support for streaming data and additional compliance regimes. Anathakrishan also highlighted the recently announced partnership with Internet2 that makes Globus subscriptions available via the NET+ Program, and the free subscriptions available to grant awardees of the National Science Foundation’s CC* program.

The keynote continued with Kyle Chard, research associate professor and Globus Labs Lead. Dr. Chard focused on a topical theme at the conference: how AI and ML is transforming scientific discovery, and how scientific computing will become increasingly “agentic”, with most steps of the scientific method outsourced to a collection of AI-enabled agents.

GlobusWorld 2025 hosted a diverse lineup of invited speakers spanning the gamut of research computing and data professionals, as well as researchers themselves. Brian Roland from Northwestern University, presented their enterprise implementation of Globus Connect Server on Amazon Web Services, and how they are managing the migration of a research data archive from on-premise storage to AWS S3 at their institution. Rick Wagner, CTO of the San Diego Supercomputing Center (SDSC), talked about the deployment of Globus Compute multi-user endpoints on SDSC’s Expanse system, and how the service can standardize programmatic access across SDSC’s HPC resources, thereby enhancing security, access modalities, integration and automation.

Dimitrios Bellos, a research software engineer from the Rosalind Franklin Institute (RFI) in the UK was awarded this year’s Globus Community award for his expansive work with the Globus platform and championing its broad deployment as critical RFI technology.

Bellos commented, “I am deeply honored and grateful to receive this year’s Globus Community Award! This award means a lot not only to me but the whole Advanced Research Computing (ARC) team here at the Rosalind Franklin Institute and we truly appreciate the recognition. The hard work that it represents is part of a team effort between Franklin’s ARC team and the Globus Support team.”

Day one ended with an outdoor reception on the Gwen Hotel terrace, where attendees could recharge and make new connections with peers and Globus team members.

Day two included a morning of sessions on transforming scientific discovery with AI and ML. Dr. Chard kicked off the morning with a presentation on how Globus supports AI processes across various stages of model training and inference, highlighting its role in facilitating advanced AI research and deployment. This was followed by invited talks on leading edge developments in AI/ML for Science: Zhao Zang, Assistant Professor at Rutgers, discussed Diamond, a service designed to facilitate large model training; Ravi Madduri, a senior scientist at Argonne National Laboratory, presented the APPFLx platform for secure, privacy-preserving, federated learning across distributed large-scale, high-performance computational resources; and Ryan Jacobs, a scientist at the University of Wisconsin-Madison, highlighted the need for the availability of persistent, easily accessible and useable machine learning models of materials.

Both days included well-attended sessions of Globus Office Hours—everyone’s favorite time, where Globus developers dive in and address attendees’ questions about the service.

The full program and list of speakers can be found here: https://www.globusworld.org/conf/program