Flora Esposito

At KU Leuven, I build hybrid thermodynamic and machine-learning models for perfume raw materials, fabric softeners, and fragrance formulations in collaboration with P&G. My work combines COSMO-RS, QSPR, and serial regression to predict activity coefficients across laundry touchpoints under scarce-data conditions.

I developed optimisation frameworks for Eau de Parfum and Eau de Toilette formulations that balance olfactory intensity, fragrance-pyramid constraints, character matching, and material cost. I also created GUI and interactive dashboard tools using tkinter, Streamlit, and Plotly to make formulation results actionable.

Previously, at INRS, I built a 3D multiscale CFD model of an industrial ammonia adsorption process in COMSOL. I contribute through publications, conference presentations, teaching chemical engineers Python and hybrid modelling, and supervising research on CO2 absorption modelling and optimisation.