Thomas Delacroix

Data Scientist with experience in machine learning, statistical analysis, and large-scale data analytics across MIT, SWISS, and Merck KGaA. Strong quantitative background in causal inference, forecasting, machine learning, and optimization with hands-on Python and SQL experience translating complex data into actionable decisions.

Operations Research / Data Science Intern at SWISS Air Lines (2025-11 – 2026-04)

• Built data pipelines to preprocess, validate, combine, and analyze large-scale operational data used for planning and decision-making.
• Developed quantitative and optimization approaches to support recovery and planning decisions in a large-scale transportation network.
• Partnered with operational stakeholders to translate analytical findings into actionable recommendations and evaluate alternative strategies.

Master's Thesis: ML-Enhanced Optimization for Supply Chain Network Design at MIT (2025-03 – 2025-09)

Intelligent Logistics Systems, Center for Transportation and Logistics

• Developed an end-to-end ML pipeline using Graph Neural Networks to predict decisions in large-scale network design problems.
• Generated and processed experimental datasets, engineered features, trained classification models, and conducted validation, benchmarking, and error analysis.
• Integrated GNN predictions into a MILP optimization pipeline, reducing solve time by 60–70% while keeping solutions within 2% of optimality on average.

Supply Chain Planner Intern at Merck KGaA (2024-09 – 2025-03)

• Analyzed demand, inventory, production, and capacity data to monitor operational KPIs, identify supply risks, and investigate root causes of performance deviations.
• Built automated analytical workflows and dashboards using Python, SAP, Palantir Foundry, and Excel, improving recurring monitoring and decision support.
• Worked with Production, Quality, Planning, Procurement, and Logistics teams to translate quantitative findings into operational actions.

Project: Social Media Sentiment & Polarization Analysis at EPFL (2024-02 – 2024-07)

• Analyzed 400K+ social media posts to investigate sentiment dynamics, political polarization, engagement patterns, and shifts in online behavior.
• Applied NLP, change-point detection, anomaly detection, and data visualization to identify temporal and behavioral patterns in large-scale user-generated data.

Project: Measuring the Impact of Cinema on U.S. Naming Trends at EPFL (2023-02 – 2023-07)

• Combined movie and U.S. baby-name datasets and developed a regression-based influence metric to quantify shifts in naming trends following movie releases.
• Used exact matching to control for observed confounders and compare naming outcomes across movies with different popularity levels and character attributes.
• Applied hypothesis testing, confidence intervals, and difference-in-differences analyses to investigate movie and character effects, communicating findings through an interactive data story.

Master's in Business Analytics – EPFL (2022-09 – 2025-10)

Bachelor in Mechanical Engineering – EPFL (2019-09 – 2022-06)