Victor F.

Data Scientist end-to-end: from data architecture to production deployment, with a focus on ROI.

For 2 years, I was the sole Data Scientist at a Swiss insurance broker (EMGSA, Geneva). I managed the entire project cycle autonomously: needs identification with management, development, AWS deployment, and production maintenance. Concrete result: 150k+ CHF in quantified annual savings.

What I deliver:

→ ML models in production — Lead scoring (CatBoost, 2.2x conversion rate improvement), behavioral prediction, anomaly detection. With automated retraining pipelines and performance monitoring.

→ LLM & agentic systems — RAG (Ollama, pgvector), MCP (FastMCP), Pydantic orchestration. I designed an agentic system for automated contractual data extraction.

→ Data architecture from A to Z — BigQuery Data Lake, ETL pipelines, feature engineering from CRM data. AWS infrastructure (EC2/S3), Docker, CI/CD.

→ Business impact automation — Scraping pipelines (Python/Selenium) processing 100+ leads/day across 8 websites, OCR + LLM for document processing.

Stack: Python • CatBoost/XGBoost/PyTorch • BigQuery • PostgreSQL • AWS • Docker • RAG/MCP • Streamlit • PySpark • Airflow

Available for remote, hybrid, or on-site work — Lyon, Geneva, Rhône-Alpes region.