When J. “Robert” Michael started his new job at the National Center for Computational Sciences (NCCS) at the US Department of Energy’s (DOE’s) Oak Ridge National Laboratory (ORNL) last September, he took on a unique task: making sure that data is a “first-class citizen” in the scientific research being conducted on NCCS supercomputers.
As the NCCS’s first designated chief data architect, Michael supports the development, refinement, and architecture of strategies to overcome challenges encountered in data-intensive computing across many NCCS programs.
What does that mean?
Traditionally, scientific research on high-performance computing (HPC) systems has used ab initio (“from first principles”) mathematical calculations to model phenomena and then perform simulations to achieve results that are contrasted with experiments. A newer approach analyzes large data models on a particular subject to extract findings, so data is the “ground truth” for the research rather than equations. Both techniques are vital tools in computational science, although they’re often used separately.
“What I’m trying to help facilitate is integrating these two approaches. We have a lot of history solving scientific problems, but I think we’re still building the expertise, the manpower, and the infrastructure to connect these systems,” Michael said. “So it’s less about convincing people that data needs to be a first-class citizen and more about helping to provide the infrastructure to allow it to be a first-class citizen.”
Robert is the NCCS’ new chief data architect
One of Michael’s first projects has been leading the Scalable Protected Infrastructure team, which is establishing security protocols for handling sensitive data that Oak Ridge Leadership Computing (OLCF) supercomputers have not previously been authorized to analyze efficiently at scale, such as patient records protected by the federal law restricting release of medical information. This required coordinating personnel from several different departments: HPC Cybersecurity and Information Engineering, HPC Scalable Systems, Scalable Protected Data, and the Information Technology Services Division.
“A lot of what I bring to the table is being able to bring together the policymakers, the program managers, the engineers, and the scientists. I’m able to kind of speak all of those languages,” Michael said. “And that’s one thing that interested me in this position of chief data architect—it fit all of those areas that I’ve enjoyed in my career, from the computational side of things to the data side of things, as well as leadership and management.”
Michael is applying this cross-team approach to a variety of programs within the NCCS. He is currently the principal investigator for the Clinical Concept Repository project, which seeks to design a scalable framework for browsing a library of clinical concepts in collaboration with the US Department of Veterans Affairs. He also serves as the biosciences lead for ORNL’s edge computing strategy, which is developing ways to process data right at the scientific instruments, as well as centralized computers. This is closely tied to his involvement with ORNL’s Compute and Data Environment for Science, which helps coordinate compute and data solutions across the lab.
Michael’s education and employment history make him uniquely suited for such interdepartmental collaborations; he has a PhD in computational science with an emphasis in quantum physics and master’s degrees in theoretical mathematics and computer science. At St. Jude Children’s Research Hospital in Memphis, he managed a team of engineers developing bioinformatics software for hybrid HPC and cloud architectures.
Although Michael now lives in Knoxville near the ORNL campus, he still works remotely from his home while the lab continues to cope with the COVID-19 pandemic. This has made team building and project management a bit more challenging, even when everyone’s cameras are turned on during video meetings.
“As the chief data architect, one thing that sets me apart from a data engineer is really the strategy piece of it. I think that’s probably one of the hardest things, especially working in a virtual world. It’s a bit easier to get to know people in person,” Michael said, “so one of the first things I set out to do was really understand the landscape and really understand what are the projects that are happening, who are the people, where are the needs.”
One overall need that Michael would like to fulfill is the development of a data management ecosystem, making sure that the data itself is tracked. This will become even more important as data science advances along with the ongoing increases in computational power, such as the OLCF’s upcoming exascale-class supercomputer, Frontier.
“In my mind, Summit and Frontier and whatever big HPC system comes next are all pieces to this large ecosystem puzzle,” Michael said. “But I have been less focused on the big HPC systems and more focused on how we integrate all of these disparate systems and do federated computing and federated data analysis. How do we do real-time analytics with very big data right there at the instrument?”
UT-Battelle LLC manages Oak Ridge National Laboratory for DOE’s Office of Science, the single largest supporter of basic research in the physical sciences in the United States. DOE’s Office of Science is working to address some of the most pressing challenges of our time. For more information, visit https://energy.gov/science.
