When scientists detected gravitational waves for the first time in 2016, confirming a century-old prediction by Einstein, the breakthrough relied on more than just sophisticated instruments. Behind the scenes, software developed at USC helped manage the massive computational effort required to identify the signal hidden in the data.

That research, and decades of related innovations, have now earned Ewa Deelman, research director at USC’s Information Sciences Institute, the 2025 IEEE Computer Society Sidney Fernbach Memorial Award, one of the field’s highest honors.

The award recognizes Deelman “for pioneering research and software engineering in distributed systems in support of scientific workflows”, work that has fundamentally changed how scientists across disciplines conduct complex computational research.

The Sidney Fernbach Memorial Award, which includes a certificate and $2,000 honorarium, specifically honors outstanding contributions in the application of high-performance computers using innovative approaches. Deelman received the award at the SC25 Conference in St. Louis on November 18, with an invited presentation on November 19.

Transforming Scientific Computing

At the heart of Deelman’s contributions are algorithms and innovations embodied in the Pegasus Workflow Management System, software that has enabled breakthrough discoveries ranging from the first direct detection of gravitational waves to mapping earthquake hazards across California.

“I am very honored to receive this award,” Deelman said. “I have spent most of my career, including my PhD, on developing algorithms and systems that support and optimize the execution of applications on high-performance and distributed systems.”

Scientific workflows, as Deelman explains, function like recipes for computational research. “The ingredients are the data that needs to be used as input and the various tasks that operate on the data and the order in which they need to occur,” she said. These workflows allow scientists to declaratively specify complex computations without worrying about the underlying computing infrastructure.

A Key Innovation: Separation of Description from Execution

Deelman identifies the separation of workflow description from workflow execution as her foundational contribution. This architectural decision enables scientists to focus on the logic of their computations rather than the technical details of the systems running them.

“This separation led to new algorithms for job scheduling, workflow optimization from the point of view of performance and data footprint, and others,” she explained. The workflow management system handles the complex task of mapping high-level scientific descriptions to executable forms tailored for specific computing environments and their execution.

The technical challenges were substantial. Deelman’s team had to develop solutions for submitting jobs across various distributed resources, managing file transfers between them, tracking provenance and performance information, and providing debugging tools. Pegasus addresses everything from automated data staging to intelligent cleanup of intermediate results, from job clustering for efficiency to sophisticated failure recovery mechanisms.

Enabling Scientific Breakthroughs

The impact of Deelman’s work spans multiple scientific domains. The Statewide California Earthquake Center uses Pegasus-managed workflows to generate sophisticated, physics-based seismic hazard maps that inform building codes throughout California, helping make structures more resilient during earthquakes.

In physics, Pegasus played a crucial role in LIGO’s historic 2016 detection of gravitational waves, the first direct observation of these ripples in spacetime predicted by Einstein. The system continues to support the Event Horizon Telescope’s black hole research, running simulations that help determine the characteristics of these mysterious cosmic objects.

Astronomers use the Montage workflows for their own research. Researchers in bioinformatics, astronomy, and numerous other fields rely on Pegasus to manage their computational pipelines.

A Career Spanning Computing Eras

Deelman’s journey began during her PhD at Rensselaer Polytechnic Institute, where she worked with a biologist on simulating the spread of Lyme disease. “I developed parallel discrete event simulation techniques to enable the simulation of the movement of mice and deer and the tick lifecycle,” she recalled. “From then on I was always interested in seeing how CS can help solve domain-specific problems.”

The early days of Pegasus coincided with the era of grid computing; the first three letters stand for “Planning for Execution in Grids.” The team initially explored AI planners for workflow creation, though these formal frameworks ultimately lacked the scalability needed for real-world scientific applications. Pegasus’s innovation was retaining the planning abstraction while developing heuristic-based approaches that could handle distributed, data-intensive workloads.

Looking to the Future

Now, Deelman is bringing her work full circle with AI. Through the NSF-funded Pegasus AI project, her team aims to enhance workflow composition, monitoring, and debugging while making systems more resilient to failures. A separate DOE-funded project called SWARM is reimagining workflow management as a fully distributed and resilient agentic solution.

In her SC25 presentation, titled “Abstraction and Automation: From Workflows to Intelligent Systems and the Future of Scientific Discovery,” Deelman will explore how AI is redefining scientific computing while emphasizing the continued importance of human judgment and critical thinking.

“Scientific computing has evolved through successive layers of abstraction, from machine code to workflows, from manual scheduling to autonomous orchestration,” she notes. As systems increasingly reason, predict, and adapt autonomously, she believes researchers must confront deeper questions about transparency, trust, and the proper balance between automation and human insight.

For Deelman, receiving this recognition validates decades of work at the intersection of computer science and scientific discovery, work that continues to evolve with each technological revolution.

Published on November 19th, 2025

Last updated on November 19th, 2025