Cyril N.

Design and deployment of advanced demand forecasting models for Global Travel Retail, aiming to predict sales volumes (bottles) per airport, airport cluster, product category, and Maison, in a context of high seasonality and dependence on passenger traffic.

Development of multi-level forecasting architectures combining time series and machine learning, ensuring consistency between global (region/cluster) and local (airport/point of sale) forecasts, to support revenue management, stock allocation, and commercial planning decisions. Integration of key exogenous factors (air traffic, tourist seasonality, events, promotional history, passenger mix) to improve forecast robustness and anticipate stock-out or overstock risks.

Design and implementation of Marketing Mix Modeling (MMM) models for Moët & Chandon and Hennessy in the US market (the group's main market), to measure the incremental impact of media investments and optimize budget allocation between Maison × Market × Channel.

Implementation of a "Smart Execution" framework facilitating the operational activation of model recommendations (forecasting, MMM, pricing), with performance tracking on key KPIs (volume, revenue, CTR, VTR).

Leading a multidisciplinary team of data scientists for the development and deployment of a SaaS solution dedicated to marketing budget optimization and revenue management.

Design and implementation of Bayesian Marketing Mix Modeling (MMM) models (PyMC, Lightweight-MMM) to optimize media investments (TV, digital, print) across several international markets.

Development of a budget allocation optimizer based on linear programming (Gurobi), integrating business and operational constraints, deployed in over 20 countries.

Integration of out-of-stock models to consider supply chain constraints in marketing decisions and maximize the real impact of investments.

Design of pricing and promotion models to support price adjustment strategies in an inflationary context and support data-driven decisions with high business stakes.

Solutions deployed on AWS cloud environments with data exploitation via Snowflake, in a large-scale production context.

⸻

Development of customer acquisition models dedicated to the activation and optimization of Facebook Lookalike audiences, using behavioral, transactional, and customer value (LTV) signals, to improve the quality of acquired users and campaign ROI.

Design and production deployment of Customer Lifetime Value (CLV) models to manage in-app offer targeting, ensure revenue forecast reliability, and guide acquisition strategies towards higher-value profiles.

Implementation of a multi-touch attribution framework combining online channels (Facebook, Google, TikTok) and offline channels (TV, sponsorship), enabling advanced marketing performance measurement and more efficient acquisition budget allocation.

Development of a budget optimization framework to dynamically arbitrate marketing investments across channels and maximize overall ROI.

Implementation of a "Next Best Offer / Next Best Action" recommender engine, aiming to recommend the best cashback offer based on recent transactions, spending profile, and user behavior, with the goal of increasing engagement and LTV uplift.

Design of customer segmentation models to identify high-value users and support targeted subscription, incentive, and retention strategies.

Solutions deployed in a fintech environment on Google Cloud Platform (GCP), in a large-scale production context.