Hippolyte D.

→ Industrialization of data pipelines for various carriers. Process automation to reduce manual workload and increase productivity (ingestion, reporting, and monitoring). Design and development of an internal geographic normalization API consumed by several teams, resulting in -99% errors and continuous repository enrichment.

→ Technical scoping for migration to a Data Warehouse, led with the Data Director: documentation writing, user training on new tools, and Python skill development for the team.

Tools: Microsoft Azure (SQL Database) – Python (Polars, FastAPI) – Power BI – Confluence

→ Redesign of architecture and migration of a data visualization application to a modern stack with integration of a machine learning prediction module: development, documentation, and production deployment.

→ Data POCs: development and production deployment of machine learning models (CNN, SVM, XGBoost).

Tools: Python (Pandas, FastAPI, scikit-learn, PyTorch, Dash Plotly) – PostgreSQL – Spark – Docker

→ Continuation of research on algorithmic performance in computer vision

✅ Improvement of all previous results (+8% overall)

✅ Implementation of scientific articles (GradCAM, transfer learning, CNN)

→ Scientific popularization POC: transfer of emotion from a real person to a portrait painting, modified accordingly (GAN)

Python (Tensorflow, PyTorch, ScikitLearn)