GeoAzur: AI for Rain & Wind from Deep-Sea Hydro-Acoustic Data
Description
Jan.2025 – Sep.2025
MSc 2 research internship at GeoAzur (Université Côte d'Azur / CNRS lab). Sole researcher on an AI approach to detect and estimate wind and rainfall over the oceans from deep-sea hydro-acoustic data (MBARI hydrophone archive on AWS S3). I designed the full pipeline from scratch: large-scale acoustic data handling, signal processing and feature extraction (MFCCs, spectral energy bands), correlation analysis against meteorological records, handling class imbalance and disjoint temporal distribution of rare rain/wind events, model development and evaluation, and a feedback learning loop for continual improvement.