Deep-sea ecosystems are poorly understood, yet monitoring them is vital to address the impacts of climate change, pollution, and noise. While passive acoustic monitoring offers a non-invasive solution, current analytical methods are manual, species-specific, and fail to scale.

​My research develops deep learning models to detect and classify marine species. Marine soundscapes are computationally challenging: signals exhibit wide bandwidths alongside extreme temporal sparsity, compounded by background oceanographic noise and high intra- and inter-species vocal variability. This project aims to build scalable models capable of learning robust acoustic representations across diverse habitats.