Nicolas Desbordes

Belfius is a company operating in the Banking and Insurance sector. The Customer Data Analytics department aims to analyze and leverage Belfius customer data. Determining the different savings modalities of Belfius customers is one of the objectives of this mission. It is therefore necessary to conduct a data study to gain a better business insight into savers on one hand, and to propose a new segmentation that highlights the specific criteria and behaviors associated with different ways of saving on the other hand.

- Research and conversion of information from raw data into a format that allows simple analysis identifying trends in different savings methods of Belfius customers

- Training of an unsupervised classification algorithm to determine a segmentation highlighting different types of savers

- Use of NLP to classify transactions based on their labels

Itelis requires the implementation of a system to detect fraudulent files for optical equipment. This involves analyzing Itelis data to determine factors that can influence fraud and thus implement a decision-making algorithm.

- POC and feasibility study of a machine-learning project on Itelis data for a fraud detection project

- Feature engineering and training of an ML model

- Production deployment of the model via Django API

Geolia Leasing Solutions is a new unregulated European player specializing in leasing and sales financing backed by a powerful investment fund. To better detect files likely to default, the risk department needs to backtest its decision model, which is a combination of business rules.

- Inspection of the leasing decision engine to identify faulty rules

- Improvement of the engine's performance through parameter calibration and rule creation/deletion