RSE-HPC-2020 Workshop Report
Published: Feb 2, 2021
We were excited to hold the inaugural Research Software Engineers in HPC Workshop (RSE-HPC-2020) as part of the SC20 conference this year! The half-day, virtual event featured 18 speakers and over 100 participants.
The opening session, chaired by Charles Ferenbaugh, began with keynote addresses from Simon Hettrick in the UK and Frank Löffler in Germany. They spoke on the growth of the RSE movements in their respective countries. Though the name “Research Software Engineer” is only about eight years old, the role is much older. In those countries (and many others), when RSE events were first organized, many RSEs joined in right away and communities grew quickly. Many who joined said they were excited to have “found their people.”
Following the keynotes, three speakers gave lightning talks on their RSE work in various contexts: Miranda Mundt on how research software engineers can support scientific software, Emilio Carcamo on the Research Software Engineering ecosystem for scientific computing supported by SC3UIS in Colombia, and Bernadette Boscoe on RSEs as maintainers and their role in keeping code alive.
We continued with two panel discussions; the first of these was on the topic of building RSE teams and groups, chaired by Daniel S. Katz with Reed Milewicz, Jeremy Cohen, Robert Sinkovits, Mahmood Shad, and Ben van Werkhoven. Several panelists described RSE groups that have developed at their institutions. In many cases the groups were informal at first, with software developers coming together to interact across different domains; later they grew into more formal groups with management agreeing to provide support. Team-building within groups is an important factor, happening both by technical means such as holding group meetings and sharing knowledge, and by non-technical means such as food and social events. Funding for groups is often a challenge. Many groups run primarily on soft money, through short- and medium-term projects; this can be effective at a large institution with many small projects in flight, but carries more risk at smaller institutions. All of the panelists were optimistic about the interesting work available to RSEs, and their growing role in the research community.
The second panel, chaired by Sandra Gesing with Joanna Leng, Ian Cosden, Rinku Gupta, and Carlos Maltzahn, discussed how to support RSE careers. Some institutions have started to recognize RSE roles, but that recognition is uneven across the research landscape. Many RSEs in academia and national labs leave for industry; this makes retention a challenge, particularly since universities can never compete on salary alone. Universities can, however, offer intangibles such as job satisfaction, flexibility, and a creative community. Many RSEs struggle to learn the skills they need for their work; some training programs and resources exist, but often the learning path is ad hoc. Mentors and peers are a great help for learning. Recognition for RSEs is sometimes difficult to gain; more efforts are needed to convince PIs and administrators that software is important. Even with the difficulties, the panelists agreed that the RSE career is interesting and rewarding.
Overall the discussions were lively and many good questions were asked. We’re grateful to everyone who took part. Slides from some of the talks are available on the agenda page. In addition, we are already making plans to (hopefully) hold a second workshop this fall at SC21 in St. Louis, MO (and/or perhaps virtually), and we hope that many of you will be able to join us!