About
I like models best once they’ve been opened up.
I study artificial intelligence at Université PSL, in the IASD master’s programme, and work as an AI engineer on a work-study contract. Before that I did a double bachelor’s in AI and economics at Paris-Dauphine.
What draws me to machine learning is usually the part after the training loop: how a GAN’s discriminator can filter what the generator makes, how decoding changes what a language model says, and whether an explanation survives when you delete the words it points to. I tend to prefer small models that can be read.
On the side, I’ve done visual-identity and graphic-design work as a volunteer, for ALIAS (2024–2025) and earlier through France Bénévolat (2020). This site is drawn in code for the same reason.
Tools & methods
- Languages
- Python, C, SQL, R
- Libraries
- PyTorch, PyTorch Lightning, scikit-learn, NumPy, pandas, Hugging Face, Captum, XGBoost
- Architectures
- Convolutional networks, Encoder–decoders, Generative adversarial networks, Recurrent networks, LSTM, Transformers
- Practice
- MLflow, Hydra, Docker, Git, Conda, uv, Linux & shell, Argo Workflows
- Mathematics
- Statistics & hypothesis testing, Bayesian methods, Time series, Linear programming, Optimisation