About me

I recently joined the Institute of Mathematics of EPFL as a Lecturer in Statistics, in Lausanne, Switzerland. I hold a Ph.D. in Statistics, which I completed in Prof. Sebastian Engelke's research group at the University of Geneva, including a stay at Columbia University (New York).

My main line of scientific research consists in developing novel methodologies for flexible conditional risk forecasting, by bridging the gap between the fields of extreme value statistics and machine learning. Other related research projects of mine involve, for example, causal modelling of extreme values, climate and weather applications, and forecasting floods. My ongoing projects focus on the links between conformal prediction and extreme value statistics, allowing high-confidence predictions. Feel free to reach out, if you are interested in discussing any of these topics!

I currently teach both large BSc-level introductory classes for various engineering programmes, and specialized MSc courses in the Statistics programme of EPFL's Institute of Mathematics. My past teaching experience includes lecturing the core BSc course Probabilités (370 students) at the University of Geneva. There, I also co-created the Advanced Topics in Machine Learning and Artificial Intelligence Master programme course about natural language processing (e.g. GPT LLMs) and reinforcement learning (e.g. autonomous robots), and assisted in teaching other classes about probability, statistics and AI during my Ph.D..

Prior to my Ph.D., I obtained a Bachelor of Science in Mathematics, a Master of Science in Applied Mathematics and a supplementary Minor in Computational Science and Engineering from EPFL, in Switzerland. During the latter two, I mainly studied, through advanced classes and projects, the topics of machine learning and deep learning, applied statistics and probability, algorithms and programming, discrete and combinatorial mathematics, and graph theory.

Current Research Interests and Projects

• Risk and rare/extreme event forecasting using machine learning and extreme value statistics
• Prediction intervals and conformal regression
• Extreme quantile regression
• Assessment and diagnostics of deep learning models for weather forecasting
• Causal modelling of extreme events

Affiliations, Memberships and Responsibilities

Current

• Lecturer in Statistics,
  Institute of Mathematics,
  EPFL, Lausanne, Switzerland.

Recent

• Lecturer & Postdoctoral Researcher,
  Research Institute for Statistics and Information Science,
  University of Geneva, Switzerland.
• Visiting Scholar (2024),
  Department of Industrial Engineering and Operations Research,
  Columbia University, New York City, USA.
• Representative at the Executive Scientific Committee of the Western Swiss Doctoral School in Statistics and Applied Probability,
  Doctoral programme in Statistics and Applied Probability,
  Conférence Universitaire de Suisse Occidentale (CUSO), Neuchâtel, Switzerland.

Ph.D. Thesis

My Ph.D. thesis, entitled Methods for Forecasting Extreme Events with Machine Learning and Extreme Value Statistics, is available in open access, and was advised by Prof. Sebastian Engelke.

If interested, you can watch a 3-minutes theatrical vulgarisation of the aim of my thesis (in French), presented at the regional final of Ma Thèse en 180 Secondes.

Short Abstract

Extreme events such as natural disasters, financial crashes, and overloaded infrastructures or services collapsing cause severe harm and lasting consequences, especially when they strike by surprise. Providing reliable risk forecasts is crucial for early preparedness, to save lives and ecosystems, and prevent economic recessions. However, foreseeing extreme events is statistically challenging, as they are unprecedented or scarce in historical records, and have complex drivers. Existing methods generally either cannot extrapolate or are not designed for accurate forecasting. This thesis develops novel methodologies for accurately forecasting the conditional risk of extreme events and for understanding their drivers, by combining the extrapolation capabilities of extreme value statistics with the predictive versatility of machine learning and with the insightfulness of causal inference. Its contributions include practical methods for predicting extreme quantiles, high-confidence intervals, and other risk metrics, a method for causal discovery in extreme regimes under confounding, and a study of leading AI weather models during extreme events.