Working with Pascal Yiou on transfer learning for the estimation of committor functions in climate models.

• Applied transfer learning to improve heatwave forecasts, extending prediction lead time from 3 to 5 days using circulation analogs.
• Trained U-Net deep learning model in IPSL-CM6A-LR and ERA5 to predict temperature fields from circulation fields, achieving correlations above 0.95 for 1-day forecasts.
• Wrote optimized code for calculating atmospheric analogs using matrix operations, significantly reducing computation time.

Working with Léonard Monsaingeon on regularized optimal transport theory.

• Implementation of Sinkhorn algorithms, Wasserstein distances and barycenters in Python.
• Study of optimal transport theory, relaxation, and entropic regularisation.

• Managed bottling and shipment logistics.
• Assisted sampling technicians during fieldwork.
• Conducted industrial prospecting activities.