Robin Munier

(Instore Analytics Department.) Led the end-to-end design and deployment of a multi-level hierarchical time-series forecasting architecture (store × department × channel) covering ~400,000 daily turnover series across 1,700 stores in 70+ countries and multiple currencies, reducing WAPE from ~30% to ~10%, leveraging probabilistic and intermittent-demand models. Implemented via distributed PySpark pipelines with automated performance monitoring. Built robust MLOps workflows (Databricks, Airflow, CI/CD). Built and productionised an LLM-based analytical system (RAG + function calling) enabling automated KPI querying, anomaly detection, and operational signal generation.

• Nestle (Purina) Designed and deployed an end-to-end machine learning solution to predict and prevent severe workplace accidents across industrial plants. Built a weekly risk-scoring system at employee, workstation, and factory-zone level, enabling proactive safety interventions. Delivered a production-ready model with >85% predictive accuracy, interpretable risk drivers, and actionable recommendations, fully integrated into an Azure/Databricks environment.

• Luminus: Developed spatio-temporal wind speed forecasting models over Belgium, reducing MAE from a 5% target to 0.2% in production. Predictive maintenance of yaw systems: achieved an 80% reduction in production loss. Optimization of wind turbines' runtime: 150MWh saved annually by turbine (on average). • BNP Paribas Fortis: Implemented probabilistic models with calibrated uncertainty estimation to automatically classify litigations-related mails into 13 categories and extract text-data specific to each category, enabling automatic actions such as bank account seizures (reducing manual review by 99%). • Cream Consulting Leader of the Data Tribe: Led a team of 10 consultants, developing internal quantitative tools (data analysis, machine learning), and contributing to recruitment, training, and technical strategy.

• Electricity consumption forecasting, supporting working capital optimisation. • Prediction and scoring of insurable events occurring in hydroelectric plants for risk quantification.