Abstract

Researchers and students entering the academic world are landing in an environment unlike any we have seen before. The generalized publish or perish policy has generated a data reproducibility crisis that threatens our credibility. On the other hand, the increase of our data storage and computational capacities, as well as the development of artificial intelligence opens huge opportunities. Instead of limited case studies, we can now compare simultaneously thousands of systems and understand the general laws governing the performances of our applications faster than we ever did. And this progress is much needed in regards in the challenges – global warming, major pandemic risks, etc. – that we are currently facing.

However, such large-scale studies require data to be structured, standardized, shared and perfectly documented, so that it can be used across projects and across institutes and contain all the information needed to ensure reproducibility. In the era of terabyte storage capacities, producing such data is no technical issue, but we need to adapt our best-practices to unlock data potential. The concept of FAIR data, for Findable, Accessible, Inter-operable and Reusable (or also Findable and AI-Ready) summarizes what we need our experimental, theoretical and simulation results to be. I will first present this concept, our need for it, and briefly the changes we need to implement to really enter the 21st century data production era. We will then explore the work of the FAIRmat consortium, one of the groups actively working on this transition, as well as NOMAD, one of the informatics tools to format, store and handle those data.

Biography

Dr. Julien Gorenflot is a research scientist in charge of implementing FAIR (Findable Accessible Interoperable Reusable) data in the division of Physical Science and Engineering in KAUST. 15 years of research in the field of photovoltaics made him realize how powerful, yet how impossible, it is to assemble data across projects, across laboratories, across universities. As a matter of fact results can barely be re-used past their initial publication, a major obstacle to their transfer to industrial applications. Other critical problems are the lack of verifiability of results, fueling the current reproducibility and retraction crisis and a gradual lost of trust in science. Realizing that this all comes down to a lack of consistency in data collection and transmission, he is now dedicating himself to defining standards and formats in close collaboration with the FAIRmat consortium in Germany, implementing them at the heart of experimental setups with the help of their users, training users and raising awareness. Bibliographic bullet points: - July 2024: starting a new life as a FAIR data implementer. - February 2016: joined KAUST as a postdoc then research scientist. - 2015: postdoctoral fellow in the Max Planck Institute for Polymer Research, Mainz, Germany. - July 2015: PhD from the Faculty of Physics and Astronimy of Würzburg, Germany. - Until 2008: life and studies in France: Meaux, Mende, Sarlat-la-Caneda, Pau, Strasbourg, Toulouse - Easter 1981: born in France, on a hill close to Paris, grew up with the Eiffel tower light at the Western window and the Sleeping Beauty Castle (Eurodisneyland park) at the Eastern window.