My primary interest is developing signal processing tools for neural time-series recordings, considering their high dimensionality, inherent stochasticity and almost inseperable noise. More specifically, my aim is to design innovative parametric/semi-parametric modeling paradigms of neural dynamics, apply machine learning (optimization) techniques to learn those model parameters in a data-driven way, and demonstrating the types of inferences (including causal ones) that can made from those model parameters. These efforts are directed towards improving neural data analysis practices with “let the data speak for itself” approach. In parallel to these technical aspects of neuro-imaging data analysis, I partake in open science and reproducible research by creating easy to use software tools and disseminating them as public repositories under open-source licenses.