David Krasowska and Kirill Nagaitsev Awarded DOE Computational Science Graduate Fellowships

Northwestern Engineering computer science PhD students David Krasowska and Kirill Nagaitsev have been awarded Department of Energy (DOE) Computational Science Graduate Fellowships (CSGF). Members of the Prescience Lab, both students are advised by Peter Dinda, professor of computer science and (by courtesy) electrical and computer engineering at the McCormick School of Engineering.

Established in 1991 for doctoral students pursuing fields that use high-performance computing to solve complex science and engineering problems, the DOE CSGF provides up to four years of total support, including a yearly stipend of $45,000, payment of full tuition and fees, and an annual $1,000 professional development allowance.

Fellows gain cross-disciplinary, team-based experience during a 12-week practicum at one of 21 DOE national laboratories or sites, including access to DOE supercomputers. They also engage in a rigorous program of study that emphasizes academic breadth. At the DOE CSGF annual program review held in Washington, DC, current and former fellows and university and DOE laboratory staff connect through outgoing fellow talks, a poster session, and professional development panels.

David Krasowska

David KrasowskaKrasowska is a first-year computer science PhD student in the computer engineering track.

His research focuses on designing processing in memory (PIM) systems for high performance computing (HPC) infrastructures like data centers. Krasowska’s goal is to develop a software approach that will take high level programming languages popular within a specific domain — such as SQL for data centers — and automatically detect parallel code to generate PIM instructions without the need for programmer intervention. He also aims to improve hardware design to optimize performance and require less energy consumption.

Krasowska is drawn to the opportunities offered by the DOE CSGF practicum.

“I look forward to making use of the computational resources that are made available by the DOE,” Krasowska said. “Hardware simulation is a computationally demanding application, and it would be a great benefit to use the extra resources.”

Krasowska is a visiting research assistant at the DOE Argonne National Laboratory. He collaborates with computational statisticians and computer scientists to investigate lossy compression for optimizations in HPC scientific applications.

He earned a bachelor’s degree in computer engineering from Clemson University, where he was advised by Jon C. Calhoun, assistant professor of electrical and computer engineering, and director of the Future Technologies in Heterogeneous and Parallel Computing Laboratory.

"I am so grateful to have received such a competitive fellowship. This is an opportunity to work with great people to gain different perspectives and showcase new and exciting research,” Krasowska said. “I would like to thank my mentors Jon C. Calhoun and Peter Dinda, and my team at Argonne National Laboratory.”

Kirill Nagaitsev

Kirill NagaitsevNagaitsev is a first-year PhD student in computer science. His research interests center around operating system and compiler support for parallel computing. He also investigates parallel programming models that can effectively utilize heterogeneous hardware.

Nagaitsev’s goal for the fellowship is to investigate how parallel programming models can be improved to allow programmers to write critical portions of their application in a high-level, parallelizable way, where they avoid getting caught up in the details of the hardware that they will be using.

“There has always been a tradeoff in programming where one could write a very flexible application that is somewhat performant on any hardware, but it likely will not perform as well as an application that is tailored to a specific piece of hardware,” Nagaitsev said. “This effort of tailoring applications to hardware is difficult, time consuming, and not future-proof, as a better piece of hardware could come along that makes the effort obsolete.”

Nagaitsev aims to push the performance boundary that is achievable for flexible programming models that can target any hardware. His research could also lead to the development of a novel middle ground between flexible, hardware-agnostic models and highly specialized programming models for specific hardware targets.

“Receiving this fellowship is both an honor and an acknowledgement of my work thus far,” Nagaitsev said. “It marks the beginning of a new path of research that I'm excited to go down for years to come.”

Prior to joining Northwestern, Nagaitsev earned a bachelor’s degree in computer science from the University of Chicago, where he was advised by Ian Foster, Arthur Holly Compton Distinguished Service Professor of Computer Science and a distinguished fellow and senior scientist at Argonne; and Kyle Chard, research associate professor at the University of Chicago. Nagaitsev worked on the NSF-funded project funcX, which enables scientists and programmers to easily submit large numbers of jobs to supercomputers or the cloud.

McCormick News Article