David Elbert is a Faculty Research Scientist at Johns Hopkins University with particular focus on innovation in data infrastructure to accelerate materials discovery and design.
His work centers on event-driven, data-centric approaches across many material classes including quantum materials, spall-resistant alloys, recycled polymers, scalable catalysts, armor materials, and certifiable metal additive manufacturing.
He is the Chief Data Officer of the PARADIM NSF Materials Innovation Platform and the data lead for the IMQCAM NASA Space Technology Research Institute.
David is a co-founder of the Materials Research Data Alliance (MaRDA) and leads the Materials Research Coordination Network (MaRCN), a U.S. National Science Foundtaion FAIR and Open Science Research Coordination Network collaboration between Johns Hopkins, Northwestern, Duke, Purdue, Chicago, Buffalo, and the University of Texas, El Paso.
Event-Driven Foundations of Autonomy: Laboratories and Digital Twins
Autonomous laboratories and executable digital twin environments demand more than automation—they require connected, intelligent cyberinfrastructure built on bidirectional data flow, semantic integration, and knowledge-based reasoning. In such complex environments, event-driven architecture serves as the foundational design principle. Event-driven approaches enable next-generation platforms with orchestration of complex, distributed workflows across instruments, computational models, and decision-making agents.
In this presentation, we detail our mission-driven development of an event-based approach within the AIMD-L (Artificial Intelligence for Materials Design Laboratory) and IMQCAM (Institute for Model-based Qualification and Certification of Additive Manufacturing) projects. This includes novel uses of our OpenMSIStream streaming framework, workflow orchestration through data portals, and emerging strategies for globally unique, persistent identifiers (PIDs) for physical samples. These tools enable real-time coordination of heterogeneous scientific components, paving the way for agile, model-centric experimentation.
Beyond infrastructure, a central challenge in achieving autonomy is representing knowledge in a form that supports both scientific discovery and intelligent decision-making. We address
this by adopting the graphical expression of materials data (GEMD) as a canonical, machine-readable structure to represent material histories, synthesis and processing steps, and characterization events. GEMD captures causal relationships across workflows, enabling provenance-aware, AI-ready experimentation.
As we move forward, we are exploring the integration of GEMD with semantic technologies and traditional RDF-based knowledge graphs, enabling hybrid reasoning systems that support adaptive decision-making, explainability, and closed-loop optimization. Together, these developments form a flexible, event-driven foundation for autonomous science platforms and certifiable digital twins in materials science.
