Lilya-Nada Khelid

• Marketing Mix Modeling (MMM) Objective: Optimize media investments and maximize sales for Armani, Mugler, Valentino, Azzaro, and Prada. Achievements: Developed an automated data pipeline to efficiently process and organize marketing data. Implemented models using Robyn and Prophet to evaluate the impact of media on sales. Performed statistical tests to ensure the robustness and reliability of the models. Designed an interactive dashboard to present insights, simulate optimal media mix scenarios, and maximize ROI.

• Sales Prediction in Finance Objective: Anticipate demand and support strategic decision-making. Achievements: Leveraged data to enhance demand prediction for new and recent products. Utilized Deep Learning and Machine Learning models to improve sales forecasting accuracy.

• Technical Objective: Automate processes and guarantee dashboard reliability. Maintained an interactive dashboard integrating sell-out for L'Oréal. Automated results and workflows using SQL, Vertex AI, and various GCP Conducted both business and technical checks to validate dashboard accuracy. Facilitated workshops with business teams to ensure coherence of insights and clear understanding across stakeholders.

Development of a RAG (Retrieval-Augmented Generation) System for a Technical Chatbot Objective: Enable auditors and engineers to quickly access technical information and past recommendations. Benchmarking: Analyzed three types of embedding models:

Multilingual French-specific Multi-vector approach

LLMs used: Mistral and Llama to generate specific technical questions. Results: Assessed model confidence and reliability. Validated robustness against diverse queries, including both keywords and technical questions.