Nuclear physicist and computer scientist by training, I worked 7 years at CEA (French Atomic Commission) on Monte Carlo methods and High Performance Computing. In 2019 I seized an opportunity to join the LLNL. I worked on the RADIUSS project, promoting LLNL open-source software for scientific computing, as well as on the MFEM project. My role was to improve the development workflow of open-source projects using automation and DevOps principles. I returned to France and created Woptim to keep helping scientific codes automate their development workflow.
Woptim was created to provide projects with my expertise in workflow optimization applied to development processes. I will help your team produce better code by addressing the bottlenecks in everyday coding and add crucial testing capability that might be lacking. I address specific issues with solutions coherent with the bigger picture so that my work is likely to last longer and be reused by others. Developing better code is a team effort, but not all teams can afford to spend time exploring tools and trends: Woptim is there to provide insights applicable to scientific codes running on super-computers.
I specialize in workflow optimization for scientific codes running on super-computers. Open-source software is meant to be available and open to contributions. However, HPC software often run on one-of-a-kind machines, which makes continuous testing challenging. Branching tools like GitHub and GitLab, packaging software with Spack and making sure multiple projects work well together are the kind of tasks I work on everyday. Most of this work is open-source (and documented).