Hello ! I am currently Head of Data Science at Nestlé Zone Europe, where I lead a team working on consumer modelling and applied AI: consumer embedding learning, self-supervised representation learning, and agentic AI frameworks to accelerate experimentation and decision support. Before that, I was Lead Data Scientist at Genesis, where I defined the R&D strategy for soil health models based on physicochemical and biological indicators, and coordinated multidisciplinary projects quantifying the environmental impact of agroecology and conservation agriculture. Before that, I was a Senior Lecturer at AgroParisTech, where I taught statistics (mostly Bayesian) and machine learning (classes at AgroParisTech and university Mohamed VI), and was doing research in generative models, self supervised learning, and environmental and ecological modelling (mostly using Bayesian methods, once again !). I did my PhD at CMAP in Ecole Polytechnique supervised by Éric Moulines and Arnaud Doucet from Oxford University.

You can find a version of my Curriculum Vitae here.

My research in academia focused on problems at the intersection of mathematics and computation, and in particular between classical statistical and mathematical modeling and Machine Learning tools. My PhD focused on “Novel Variational Methods for Inference and Learning in high dimensions”, and especially around Monte Carlo methods, dealing with Markov chain Monte Carlo, Variational Inference, generative modeling. I am also very interested in uncertainty quantification, using Bayesian inference. You can find a pdf version of the manuscript here.

Previously, after 2 years of prépa at Stanislas, Paris, I got into the École polytechnique in 2015, where I was in the “section rugby”. In 2018/2019, I went to Oxford University for the Master of Science in Statistical science, from which I graduated with distinction.

I am very interested nowadays in real world applications, especially in the agricultural and agronomic fields, and I keep a few side projects running on my own time:

  • Agriculture — a cluster of tools for French farmers:
    • Agri Helper: crop yield forecasts with uncertainty (Monte Carlo weather simulation + quantile regression, P10/P50/P90), plus crop-health tools such as computer-vision detection of weeds and thistles.
    • Culture recommender (AgriTransition 2050): decision support for crop-transition choices under climate change.
    • Agricultural foundation model: an early-stage self-supervised model for agricultural parcels, built on Sentinel-2 imagery, French RPG/PAC parcel data, ERA5-Land weather and SoilGrids.
  • Fermentation — from process control to knowledge tooling:
    • Fermentation control: a soft-sensor + model-predictive-control POC (mechanistic Monod kinetics combined with ML soft sensors), aimed first at wine fermentation, with precision fermentation / alternative proteins as a secondary target.
    • Fermentation digital twin & FermentGraph: a data and knowledge scaffold linking ingredients, flavour compounds and microbes across public databases, feeding an evidence-ranking workbench for fermentation R&D.
    • FermentTrack: a batch-tracking and sensor-integration app for home fermenters, standalone for now with FermentGraph integration planned; backend live, UI in progress.
  • Distributional self-supervised learning — research on joint-embedding self-supervised methods that learn distribution-valued (rather than point) representations, combining ideas from variational inference and Bayesian neural networks with modern SSL (JEPA-style architectures).
  • Personalized nutrition — an ongoing project exploring data-driven, individualized nutrition recommendations.

Always keen on hearing about opportunities and projects in those fields !

Apart from that, I play tennis, follow rugby, and do a lot of experiments about cooking, brewing, fermenting or curing anything I can get my hands on ! A blog page to document those kitchen/fermentation experiments is on its way.