My research is all about parallelizing algorithms. Some are regular, but most are irregular, i.e. non-trivially parallelizable. I use my knowledge of modern multi-, many-core and GPU architectures to rethink existing algorithms or create new ones by proposing several levels of parallelization: Instructions, cores and many-core accelarators. After working on the creation of parallel algorithms for connected component labeling, I have more recently contributed to preference learning and the computation of fuzzy landscapes.
My current teaching at CentraleSupélec focuses on parallel programming (multi-core and GPU), the discovery of C++ and the design of connected objects (from electronics to embedded computing). Particularly involved in active teaching, I favor the “learning-by-doing” approach and supervise the CROCS robotics projects, the CACS rocket and the Dassault UAV Challenge.
The ASI (Computer Systems Architecture) specialisation trains systems architects capable of understanding the technical, economic and managerial dimensions of complex industrial IT projects.