Projects
Molecular Graph Captioning — Kaggle Competition
2025- Retrieval-based graph machine learning approach to generate natural language captions from molecular graphs.
- Leveraged graph embeddings for molecule–text alignment in a chemistry-informed setting.
Frameworks / Languages: Python, PyTorch, RDKit
Research Paper Studies & Re-implementations
2024–2025 Frameworks / Languages: Python, PyTorch, TRL, transformers
Hi!ckathon (Hi!Paris) Finalist
2025- Finalist and ranked 2nd/68 (tied) in model performance within a team of 6 students
- Engineered a gated model architecture aggregating a probabilistic router (XGBoost) and a regressor expert to handle zero-inflated target distributions
- Designed a comprehensive business strategy and delivered a pitch to the final jury
Frameworks / Languages: Python, XGBoost, Scikit-learn
Cassiopée project, Télécom SudParis
2025- Assessed anonymized databases for privacy risks using OSINT methods
- Authored a comprehensive report under supervision of Maryline Laurent (Director, Networks and Telecom Dept.) and Louis-Philippe Sondeck (CEO, Clever Identity)
Frameworks / Languages: Python
TIPE: Transport Network Modeling and Optimization, Lycée Fabert
2023- Developed a mathematical and computational model of a transport network using real-world datasets
- Performed optimization and evaluated model performance
Frameworks / Languages: Python, Pandas