My research focuses on machine learning for video understanding and time series forecasting. I am experienced in self-supervised learning, compressed-domain video analysis, and graph-based sequence modeling. I develop advanced models to improve video understanding, particularly in compressed domains and through self-supervised techniques. Additionally, I explore novel methods for representing sequential data using de Bruijn graphs, including multivariate time series, protein sequences, and other temporal patterns, to enhance model performance in classification and forecasting tasks.