The Exascale Computing Project (ECP) has accelerated the delivery of a capable exascale computing ecosystem to provide breakthrough solutions for the nation’s most critical challenges. But the impacts of the project go beyond technology. The ECP has also provided tremendous opportunities for young researchers to develop knowledge and relationships within the high-performance computing (HPC) community by supporting continued individual and collaborative excellence in the field.
Tapasya Patki is a computer scientist at Lawrence Livermore National Laboratory (LLNL), the lead investigator in ECP’s Power Steering project, and the co-lead for the ECP’s Argo project. Her group’s work has greatly improved the performance of supercomputers such as El Capitan, Frontier, and Aurora by optimizing energy utilization and power management, thereby allowing the systems to complete runs more quickly and efficiently.
Patki became interested in HPC while working toward her PhD at the University of Arizona, where she developed a focus in power-constrained supercomputing and large-scale resource management. After joining LLNL as a postdoctoral researcher, she quickly rose to a leadership position within the Power Steering research group. “I had the technical expertise to do it,” she said, “but this opportunity let me apply those skills while I learned leadership, budget management, and all the other parts of running a successful team that I had less experience with at the time.”
Patki also credits the ECP with facilitating collaboration within her tight-knit research group and accelerating progress in her work. “I am very grateful for having this almost 7-year experience. ECP gave us the opportunity to streamline these processes and get much more done in much less time than we could have hoped to otherwise.” She expects to maintain a close professional relationship with her ECP colleagues and external collaborators as they begin new HPC projects.
For early-career scientists interested in HPC, Patki highly recommends attending conferences as the best way to acquire technical knowledge and make important connections with peers and mentors in the field.