Thomas Carniel developed robust data and video processing pipelines for the dataforgoodfr/13_democratiser_sobriete and dataforgoodfr/13_reveler_inegalites_cinema repositories. He engineered a Python-based OpenAlex article scraper with modular metadata extraction, PDF download, and improved CI/CD reliability using GitHub Actions and YAML configuration. For cinema trailer analysis, he built an end-to-end video evaluation pipeline integrating face detection, attribute classification, and annotated-frame outputs, with confidence-aware metrics for deeper insights. His work emphasized code maintainability through refactoring, input validation, and documentation, resulting in reproducible, reliable workflows that support richer data assets and actionable analysis in both research and production contexts.