AI/ML & Statistical Research

My research starts from a foundational principle: how do we combine rigorous mathematical statistics with modern deep learning to build models that are reliable, robust, and empirically sound? In high-stakes fields like public health and low-resource multilingual communication, a model cannot merely optimize for benchmark accuracy—it must generalize under distribution shifts and provide transparent inference.

This perspective guides my work across three interconnected areas. In Natural Language Processing, I design multi-model and ensemble frameworks to classify complex sentiment in low-resource and code-mixed settings (such as English-Bangla app reviews), addressing noisy informal text where single models often fail. In Public Health & Epidemiology, I apply biostatistical modeling and machine learning to analyze health indicators, disease trajectories, and demographic health surveys. In Statistical Machine Learning, I focus on uncertainty estimation, robust inference, and principled evaluation protocols to ensure models do not rely on spurious correlations.

Across all projects, my priority is empirical rigor and thoughtful evaluation. A model that achieves high cross-validation metrics but breaks down in real-world deployment is not practically useful. My goal is to build methods whose assumptions, limits, and predictive uncertainties are transparent and dependable.

Looking ahead, I am actively seeking Ph.D. positions in Machine Learning, AI, and Applied Statistics, particularly in research groups where rigorous methodology, open artifacts, and meaningful applications to healthcare and NLP are taken seriously.