Oscar Stanchi

I'm a researcher in Computer Science at UNLP (Universidad Nacional de La Plata) currently working on Sign Language Translation and Machine Learning interpretability. My research focuses on developing and analyzing deep learning models for sign language processing, with a particular emphasis on transformer-based architectures and attention mechanisms.

I work closely with the III-LIDI research lab team on creating datasets and models for Argentinian Sign Language (LSA) translation. My recent work has involved studying the interpretability of transformer models for Sign Language Translation, analyzing how these models process and align visual input with sequential glosses. I'm particularly interested in making sign language translation systems more transparent and reliable for real-world applications.

Beyond sign language translation, I also contribute to research in computer vision and deep learning interpretability. I've worked on implementing interpretability algorithms like RISE for the Captum framework, and participated in projects involving fundus image quality assessment in medical imaging applications. My goal is to develop more accessible and interpretable AI systems that can bridge communication gaps and enhance human-computer interaction.