2024 – 2025
- (ESILV): Predictive Modeling of Transport Equations
Transport equations refer to the evolution of a quantity over time and space, and understanding transport phenomena is essential for many industrial applications.
Based on numerical analysis, an automated system is developed to simulate and predict the evolution of transport equations.
- (Ponera) :
Anticipating Return Risk Before Product Launch
Some products sold on Amazon are returned too often, causing costs to rise. The project aims to create an AI solution that predicts products with a high risk of returns even before they go on sale online.
2023 – 2024
- (Mediarithmics/Easyence) :
Protecting Customer Data with Differential Privacy
For a machine learning model, not only should high accuracy be the goal, but privacy guarantees are also important.
The aim of the project is to use Differential Privacy in machine learning models to preserve the privacy of the training data, enabling the analysis of sensitive customer data while minimizing the risk of identifying individuals
- (Cultura/Easyence): Product Recommendation Using Image Similarity
The goal of the project is to create an effective recommender system: for a given product of interest, we would like to recommend products that look similar.
2022 – 2023
- (Décathlon/Easyence) : Faster Market Trends Detection through Machine Learning
This is anomaly detection in time series.
Create an alternative machine learning model to traditional ones that enables faster detection of changes
in the market, such as customer behavior and preferences, to identifying gaps, patterns, and opportunities.
- (Groupe Lulilo/Easyence) :
Customer Journey Optimization
Identify key touchpoints where customers interact digitally with a brand.
Create algorithms that enable delivering user experience by analyzing customer data and predicting behaviors.
2021 – 2022
- (Décathlon/Easyence): Segmenting Clients Based on Their Purchases
This is clustering. Segment customers based on their purchase behavior using clustering algorithms.
Deploy a scoring system in production to optimize marketing targeting.
- (Bricomarché/Easyence) :
Analysis of Promotional Campaign Performance
Create retail KPIs to analyze effectiveness of campaigns on sales and customer behavior.