Head of research @Califrais

I am currently head of research at Califrais. Our goal is to optimize large-scale food flows, in order to contribute to the decarbonization of the food supply chain. Our research is at the intersection of machine learning, logistics optimization, and ecology, with topics such as demand forecasting, inventory optimization, routing problems… To learn more, check out our papers here.

We are recruiting! We have

  • an offer for a 2-year post-doctorate at Sorbonne University’s Laboratoire de Probabilités, Statistique et Modélisation (LPSM) and at Califrais, co-supervised by Gérard Biau, Claire Boyer and myself. You can find the offer here.

  • an offer for a 6-8months internship, in the Califrais lab, in partnership with Sorbonne University. The goal is to study strategies for decarbonizing food flows on the French wholesale market network. You can find the offer here.

If you want to get in touch, you can contact me at adeline.fermanian@califrais.fr

Short bio

I have joined the research team of Califrais in April 2023.

Before that, I was a postdoctoral researcher at the Center for Computational Biology (CBIO), Mines ParisTech, between November 2021 and November 2022, under the supervision of Chloé-Agathe Azencott, and funded by the PRAIRIE institute.

Previously, I was a Ph.D. student in Statistics at the LPSM, Sorbonne Université, under the supervision of Gérard Biau and Benoît Cadre.

You can find my complete CV here: Curriculum Vitae,

Research interests

Currently, I am very interested in topics at the intersection of optimization and machine learning, with industrial applications in mind, the general question being how to account for uncertainty in an optimization framework. My other interests include time series and deep learning theory. The topic of my PhD thesis was the application of signatures, a tool from stochastic analysis that allows to extract information from time series, to statistics and machine learning. This topic has been developed in two directions: on the one hand, the design of new algorithms using signatures as features, and on the other hand, the use of signature theory to study existing deep learning algorithms, such as RNN, via the recent notion of neural ordinary differential equations. See the DataSig website for more information on applications of signatures in machine learning.