Scientific software development Python/HPC

Required Education : Master ou ingénieur
Start date : 22 October 2022
Mission duration : 12-18 mois
Deadline for applications : 15 January 2023
Salary : Suivant expérience

Location: Toulouse.

CERFACS provides scientific software and distributes it to its academic and industrial partners (https://cerfacs.fr/en/27112-2/). Its tools are used at the industrial (AIRBUS, SAFRAN, TOTAL) to perform high fidelity simulations on supercomputers. This is why the COOP team (https://cerfacs.fr/coop/) is interested in the process of maturing and deploying these programs. The main language targeted is Python.

A codebase rendered with gource
How can we co-develop, with dozen of authors, a preformant and high fidelity simulation software?

Description:
The candidate will evolve within a team of 4-5 people., on several projects, all trying to reduce the human work required by HPC software development and usage

Most of our projects are open-sources and codebases can be found on the Cerfacs’s gitlab.com. The work is half code development (Python, Fortran, Code Generation, sometimes C/C++), half reformulation of what was learn into the public COOP blog. The audience is a combination of Cerfacs members, Industry engineers (Safran, Airbus) and researchers from europe through our European Center of Excellence projects.

Frequent and direct contacts with the Aerodynamics / Combustion team, the SAFRAN engineers, and other COOP (Machine Learning) activities are expected.

Aim:
In 2022 our goals are four:

• improve the technology transfer between research and industry
• provide tools and techniques to make exascale computing a practical asset for simulkation-based design
• explore the young field of code metrics to bring usefull insights to HPC code developers
• create innovative training material on code development best practices

You can read more on these topics on our Team Blog.

Experience:
The candidate must have software development skills (CI, tests, lint, bloaters) and a first-hand experience in Numpy / Scipy scientific Python. He must also have a some affinities with the scientific fields covered by Cerfacs (https://cerfacs.fr/en/research-at-cerfacs/).

Contact:
We adapt a lot the work proposed to the professional project of our collaborators. You can read more about our full selection process here.

Antoine Dauptain – dauptain@cerfacs.fr

Gabriel Staffelbach – staffelbach@cerfacs.fr