Osni Marques, a staff scientist in the Applied Mathematics and Computational Research Division at Lawrence Berkeley National Laboratory (Berkeley Lab), has been tapped to lead the Training & Productivity (T&P) effort within the U.S. Department of Energy’s Exascale Computing Project. He takes over for Ashley Barker of Oak Ridge National Laboratory.

Marques has been involved with T&P for the last five years, working primarily on two initiatives:

• Advancing Software Productivity for Exascale Applications (IDEAS-ECP), which evaluates and disseminates best practices and supporting methods and tools, to improve developer productivity, software sustainability, and scientific reproducibility. He leads the Best Practices for HPC Software Developers Webinar Series (HPC-BP) in IDEAS-ECP.
• ECP Training, which consists of identifying training needs in the ECP community, responding to those needs, helping in the organization and implementation of ECP related activities, and coordination and communication of training activities in ECP and DOE computing facilities.

“As we know, software quality is a critical component of quality science, and that is one of the key things the T&P group is focused on,” said Marques, who will continue to lead the IDEAS-ECP and ECP Training initiatives in addition to his new duties as T&P lead. “I’m looking forward to working with the ECP community and facilities to identify potential software training needs going forward.”

As part of these efforts, IDEAS-ECP is already planning the transition to a post-ECP role for training and productivity. “As previous ECP reviews have noted, it is important that this work continues post-ECP, whether at the facilities or through other mechanisms,” Marques added.

At Berkeley Lab, Marques – who has been with the Lab for 26 years – has been involved in a broad range of computational research projects, including developing eigenvalue solvers for applications in engineering, biochemistry, earth sciences; electronic structure calculations and algorithms for dense linear algebra calculations in the LAPACK and ScaLAPACK libraries; and the SciDAC FASTMath Institute.

For the last 10 years, Marques has also overseen Berkeley Lab’s popular Computing Sciences Area Summer Program, which offers undergraduate and graduate students and faculty in science and engineering the opportunity to gain research experience working with staff, mentors, and peers in Berkeley Lab’s Computational Sciences Area.