About Me
I'm an engineering student at CentraleSupélec working on representation learning for sensor data. Currently, I'm co-founder at ZenkoLab, where we deploy deep learning for medical imaging. I'm also exploring how hyperbolic geometry could improve world models—research that bridges my interest in geometric deep learning with practical applications.
PyTorch JAX Scikit-Learn Computer Vision Python
Experience
Co-founder & CEO
ZenkoLab
- Engineered a pipeline handling high-dimensional, noisy retinal imaging data (fundus and OCT) in partnership with the Hopital National de 15-20, covering acquisition, cleaning, and validation across 200k+ annotated images.
- Fine-tuned a Vision Transformer (RETFound) and built a hybrid architecture (ViT, EfficientNet, Random Forest) achieving 99.6% sensitivity on diabetic retinopathy and 98.6% on AMD detection.
- Integrated Integrated Gradients, Grad-CAM, and LIME to analyze decision boundaries and failure modes in a safety-critical diagnostic setting.
Education
Master of Engineering (Diplôme d'Ingénieur)
CentraleSupélec
- Relevant coursework: Statistical Learning, Deep Learning, Probability & Measure, Optimization, Signal Processing, Data-Driven Control, Time-Series Analysis.
- GPA: 4.0/4.0
- Location: Gif-sur-Yvette, France
Global Strategy & Innovation Semester (Exchange)
ESSEC Business School
- Focus: Business Data Science, Machine Learning, Strategy, Innovation Management.
- Location: Singapore, Singapore
Preparatory Classes, MPSI/MP* (Mathematics & Physics)
Lycée Henri IV
- Rigorous two-year curriculum in advanced Mathematics (Algebra, Analysis, Probability) and Physics.
- GPA: 4.0/4.0
- Location: Paris, France