Developed srfloat: a C++ simulation library for limited-precision stochastic rounding and round to nearest in low precision, with optional Python bindings.
Research Experience
Lecturer (Maître de conférences) at Sorbonne University since September 2025; previously, postdoctoral researcher at the University of Rennes working with Olivier Sentieys and Silviu-Ioan Filip, focusing on error analysis of deep neural networks using mixed/low-precision arithmetic and stochastic rounding.
Education
PhD student at Paris-Saclay University from February 1, 2021, to December 19, 2023, under the supervision of Devan Sohier, Pablo de Oliveira Castro, and Eric Petit.
Background
Research interests include floating-point arithmetic, stochastic rounding, rounding error analysis, low and mixed precision, deep learning, quantization, and numerical linear algebra. He is a lecturer at Sorbonne University and a member of the PEQUAN team.