I am a PhD in applied probability looking for a position at the intersection of data science and communication.

(Scientific paper written with Lioudmila Vostrikova and published in "Theory of Probability and its Applications", august 2020)

In this paper, we obtained new results concerning stylized models of ruin for insurance companies investing in financial assets. The preprint version is available on HAL.

(Scientific paper written with Yuchao Dong and published in "Insurance: Mathematics and Economics", march 2020)

In this paper, we developed a new functional limit theorem similar to Donsker's theorem and the Cox-Ross-Rubinstein binomial approximation which shows that Generalized Ornstein-Uhlenbeck processes appear naturally in numerous problems in applied probability. The preprint version is available on arXiv.

(PhD thesis, defended in december 2019, supervised by Lioudmila Vostrikova)

In my thesis, I studied Generalized Ornstein-Uhlenbeck processes in the context of applied probability and, in particular insurance mathematics. The presentation slides of the defense (PDF) are available by following this link. The full thesis (PDF) is available by following this link.

(Outreach blog post, november 2020)

In november, I wrote a blog post about the mathematics behind A/B testing. After explaining the main concepts behind statistical tests, I show how Pearson's chi-squared test can be derived using only the central limit theorem. The piece was published by the influential blog "Towards Data Science".

(Python code, since september 2020)

In this project, I implemented a few useful data science functions in Python. In the first part concerning exploratory data analysis and based on mdUtils, Sklearn and XGBoost, I created a script to generate a Markdown summary file from a Pandas DataFrame and I implemented the "gradient boosting" method for linear models. The code is available on GitHub. In the second part concerning classification problems, I used the Keras-Sklearn wrapper to tune the hyper-parameters of a neural network using Sklearn's "GridSearchCV" function. The code is available on GitHub.

(Python application, since may 2020)

Note Binder is a tool to organise multiple text files in one place. It creates a database linking to the files and gives access through a simple user interface. The goal is to create an offline and movable "note book" which supports different file types and encryption. Created using Python, PyQt5 and sqlite3. The code and a macOS binary are available on GitHub.

(Series of blog posts and Python code, since march 2020)

The goal of this project is to develop a basic python framework and a tutorial to create a functional and secure Ethereum wallet. The first part has been published as a top story on hackernoon.com.

(Music and DJing project, since 2018)

Since 2018, I became interested in the creation and diffusion of techno music. Some of my mixes are available on Soundcloud and on Mixcloud.