PhD focused on approaches at the intersection of Deep Learning and Dynamical systems. The Work was conducted in collaboration with Inria and industrial partners, leading to several publications and applications. The main topics explored were :
Differential equations and time series forecasting
Machine learning
Model interpretability
Partially observed systems
Numerical simulation
2 month visit within the group of Pr Petros Koumoutsakos at Harvard
Proposed a novel, interpretable model to learn dynamical laws from data with strong theoretical links
Implementation within the numerical framework of the group.
Led to a publication, currently in preparation
Research engineer within the Computational & Data Science group.
Work on gas radiation models for the simulation of hydrogen production plants
Creation of physical system’s surrogate models using deep learning to be used within process optimization tools
Research work on the topic of Machine Learning - CFD hybridization
GPT Implementation
Leverage open LLMs to generate data
Fine tune a smaller model for portability
Use Vision Transformer for direct simulation
Pre-train then transfer on a different task