Cléophas FOURNIER

Lead Data & IA / Python – Projet Creedengo (Green Code Initiative) Création de règles d’éco‑code et coaching lors d’un hackathon sur l’éco‑conception logicielle.

Reviewed scientific literature and designed a method to estimate the environmental footprint of LLMs. Implemented an environmental footprint estimation plugin in the backend of a RAG-based application. Data classification workflows using GenAI models (Mistral 7B on GPU, Claude Sonnet via AWS Bedrock). Built a regression pipeline to predict LLM output length from input prompts, comparing an XGBoost baseline with a fine-tuned ModernBERT model. Developed 12 Python code optimization rules for SonarQube, integrated into the CI/CD pipeline. Served as Data/AI Lead for the Green Code Initiative association. Built a RAG agent to answer questions based on our responsible digital documentation.

Optimize concierge management operations through a scalable web platform: - Development of a Progressive Web App (PWA) to manage concierge activities using React and the Next.js framework - Creation of a data pipeline to collect information from multiple sources (Airbnb, Booking) - Implementation of automated workflows with Make, interacting with the application's API endpoints - Configuration of an automated CI/CD pipeline with GitHub and Vercel for continuous deployment

Process Optimization: Improved monitoring and data integration through automation, reducing processing time. Technical Development: Contributed to the development of scripts for alert automation and data tracking. Reporting Innovation: Implemented a dashboard for real-time data modification and monitoring. Significant Impact: Strengthened CITiO’s operations, enhancing urban mobility and user experience.