Data and AI Systems Section

Managing and gaining knowledge from data.

The Data and AI Systems section conducts research on the data and artificial intelligence (AI) foundations of scientific discovery. The section’s goal is to shorten and automate the path from instrument and simulation to scientific insight, and to do so in a way that scientists can trust at the scale of leadership computing facilities.

Researchers work across the full lifecycle of scientific data, from capture and curation under the FAIR (findable, accessible, interoperable, and reusable) principles, through movement, reduction, and orchestration across computing and experimental facilities, to learning, reasoning, and visual analysis under quantified uncertainty.

The Data and AI Systems section’s research agenda pursues three objectives, namely (i) advancing AI methods that are co-designed with the platforms on which they run, from experimental instruments and edge sensors to exascale systems; (ii) building the data and workflow infrastructure that couples experiment, simulation, and AI into closed loops that operate with limited human intervention; and (iii) ensuring that the resulting systems yield evidence that scientists can inspect and reproduce, through provenance, uncertainty quantification, and independent evaluation.

The team applies these capabilities both to open scientific research and to secure federal mission programs, in domains that include fusion energy, materials science, biology, astronomy, accelerator science, and national security.

The section has capabilities in the following areas:

Data Engineering | Engineering data assets and systems, including data wrangling and system design, to drive scientific innovation.
Learning Systems | Artificial Intelligence, machine learning, and data analysis at scale.
Performance Engineering | Engineering and management of secure systems.
Visualization | Methods, tools, and technologies for visual data analysis.
Workflow Systems | Large scale data management, workflow, and data compression.