Soline MOBETIE

Agentic Systems - Retail Architected and deployed a multi-agent LLM system to automate quality control for a major retail client. Built scalable GCP infrastructure and automated supplier communication workflows, reducing manual validation time by 40+ hours per operation. Propensity Scores - Banking Led the end-to-end development and deployment of product propensity scoring models for a major banking client, enabling smarter cross-sell strategies. Conducted an in-depth audit of existing models, identified key flaws in performance evaluation, and delivered a detailed report with actionable recommendations. Designed and implemented improved models, significantly boosting predictive performance and robustness, now being tested in a live pilot. Acted as technical lead on solution design and deployment, ensuring scalability, reliability, and alignment with business objectives.

Evaluated Large Language Models and built RAG-based systems for business Q&A automation. Delivered time-series forecasting models for absenteeism and supply chain optimization. Designed Bayesian marketing mix models to improve budget allocation. Defined a data strategy for the digital team of a major group. Built KPI monitoring dashboards, improving reporting efficiency across stakeholders.

Redesigned HR data models, improving efficiency and scalability. Automated data preparation workflows, reducing manual processing time. Ran A/B testing on discount strategies, increasing sales for a quick-service restaurant client. Developed deep learning and graph-based models for demand forecasting and hyper-personalization.