Jules Dubois

Led the development, deployment and scaling of demand forecasting models across nine markets using Azure, Databricks, Snowflake, dbt and Streamlit, built automated data pipelines for model training and monitoring, developed interactive dashboards for business users, conducted market-specific analyses and drived continuous alignment with cross-functional stakeholders—translating business needs into technical solutions, gathering feedback, and ensuring adoption across markets

Developed and deployed Machine Learning models for demand forecasting for the Spanish market, closely aligned with business stakeholders to ensure models address real-world constraints and deliver actionable insights

- Contributed to the Matrix Marketing Mix Modeling (MMM) initiative for the US market in collaboration with BCG, supporting model calibration, diagnostics and media ROI analysis using Bayesian models and Python-based analytics - Led the Canadian deployment of the Matrix MMM solution, implementing a new model adapted to local market dynamics, managing delivery on Azure and supervising a team of two junior Data Scientists to ensure quality and timely execution - Developed a Proof of Concept for a demand forecasting pipeline, applying time series and Machine Learning models to evaluate predictive performance and inform supply chain planning strategies

Development of Data Science solutions for Air France Industries using diverse Machine Learning techniques - Predictive maintenance on several systems from different aircraft types on Prognos software - Prediction of engine stall events on GE90 - Software maintenance on Prognos using Agile methodology and JIRA, Git, Bamboo, Control-M and AWS Development of an optimization software to plan engine removals on the whole fleet up to 4 years Management and development of a tool providing optimized plannings using Python/Gurobi Development of a model aiming at predicting the redemption rate of all Air France clients for the finance business using clustering Development of a Data science community within the company

End-of-study internship in Data Science Creating a Python/CPLEX tool to reduce turn-around time on aircrafts (Airbus A320 family) going through heavy maintenance checks (C, D, L). The tool uses the history of previous checks to predict an optimized planning capable of dealing with unexpected findings.