Artificial intelligence news.
Straight from the original announcements.

USC ISI to Present AI, Workflow and Cybersecurity Research at SC26

November 13, 2026, 22:43By the AI News Bites Editorial TeamSource: USC Viterbi (AI)
USC ISI to Present AI, Workflow and Cybersecurity Research at SC26
Image: USC Viterbi (AI)

Researchers from the University of Southern California’s Information Sciences Institute will present work on scientific workflow automation, artificial intelligence for high performance computing and cybersecurity experimentation at SC26. The International Conference for High Performance Computing, Networking, Storage, and Analysis takes place November 15, 20 in Chicago.

Key points
  • USC ISI will present research at SC26 in Chicago from November 15, 20.
  • A paper on adaptive LLM gating for HPC scheduling agents received Distinguished Paper recognition.
  • PegasusAI explores AI support across workflow development, execution and troubleshooting.
  • SPHERE will demonstrate distributed cybersecurity and privacy experimentation.

Research targets complex scientific workflows

USC ISI researchers and collaborators will present three papers at the 21st Workshop on Workflows in Support of Large Scale Science, known as WORKS26. The work addresses decentralized scheduling of scientific workflows, the use of large language models to coordinate multiple high performance computing scheduling agents, and challenges in AI assisted workflow management.

One paper, RLSA: Adaptive LLM Gating for Multi-Agent HPC Scheduling Agents, authored by Hongwei Jin, Krishnan Raghavan, Anirban Mandal, and Ewa Deelman, has received a Distinguished Paper recognition.

Scientific workflows coordinate complex computations involving multiple programs, large datasets and distributed computing resources. According to Ewa Deelman, a research professor at USC and leader of the Pegasus scientific workflow research group, researchers increasingly need to coordinate applications across different resources, manage data and recover from failures.

Deelman, USC ISI postdoctoral researcher Hamza Safri and the DOE-funded SWARM, or Scientific Workflow Applications on Resilient Metasystem, project team will present work on agentic workflow systems. Their research examines how workflows can operate across heterogeneous computing environments while responding to changing resource conditions and failures.

PegasusAI and CI Compass focus on practical adoption

In another presentation, Deelman and Komal Thareja from UNC RENCI will discuss the NSF-funded PegasusAI project. It investigates how artificial intelligence can assist scientists throughout the workflow lifecycle, including workflow development, execution and troubleshooting.

Deelman will also share updates from NSF CI Compass, the NSF Cyberinfrastructure Center of Excellence. The center provides independent technical advice, connects research facilities with the broader cyberinfrastructure community and promotes the sharing of expertise, solutions and best practices.

CI Compass helps major and midscale scientific research facilities make infrastructure decisions, leverage existing technologies and avoid duplication of effort. The project is also helping facilities examine how artificial intelligence can be adopted responsibly and effectively in operational and scientific lifecycles.

The workflow research team will meet attendees at the RENCI exhibition booth, number 1655. Presentations on SWARM, PegasusAI and NSF CI Compass are scheduled for November 17 and November 18.

SPHERE demonstrates distributed cybersecurity research

USC ISI will also participate in SC26’s Network Research Exhibition through SPHERE, the Security and Privacy Heterogeneous Environment for Reproducible Experimentation project. Led by Jelena Mirkovic, principal scientist at USC ISI and research associate professor of computer science, SPHERE provides distributed infrastructure for cybersecurity and privacy experiments using computing, networking, Internet of Things and industrial control system technologies.

Christopher Tran of USC ISI will demonstrate remotely accessible IoT devices, a miniature physical water treatment testbed with industrial control equipment, and the integration of researcher owned devices and portable facilities into distributed experiments. The demonstrations will show how geographically distributed resources can support experimentation, real time data collection and visualization.

“SPHERE makes realistic cybersecurity experimentation accessible by giving researchers access to specialized equipment and facilities that would otherwise be difficult or costly to use,” Mirkovic said in the USC Viterbi announcement. The demonstrations will take place at the SCinet Research Hotel from November 17 to November 19, with a presentation scheduled for November 18.

For users of scientific computing and cybersecurity infrastructure, the projects show how USC ISI is connecting AI assisted workflow management, resilient computing and reproducible experimentation with the facilities and communities that use these technologies.

Related news

NVIDIA cuBLAS adds FP64 algorithms for faster scientific simulations
Chips and infrastructureNovember 13, 2026, 23:13

NVIDIA cuBLAS adds FP64 algorithms for faster scientific simulations

NVIDIA has added two algorithms to cuBLAS that accelerate double-precision matrix multiplication on integer Tensor Cores while preserving full FP64 accuracy. The company…

Google details Antigravity’s multiagent results in mathematics and engineeringTop story
Products and toolsNovember 13, 2026, 21:47

Google details Antigravity’s multiagent results in mathematics and engineering

Google says its Antigravity /teamwork multiagent systems have helped human researchers address open problems across mathematics, chemistry, systems software and machine…

PegasusAI Studio Connects AI Agents to Real Scientific Computing
Products and toolsNovember 12, 2026, 21:39

PegasusAI Studio Connects AI Agents to Real Scientific Computing

PegasusAI Studio, an open-source platform from USC and research partners, is designed to help AI agents build, run, monitor and debug workflows on real scientific…