My research focuses on the algorithmic and theoretical foundations of fair and robust machine learning, with an emphasis on designing methods that remain reliable under real world constraints such as noise, bias, and structural complexity.
I work at the intersection of machine learning theory, approximation algorithms, and data driven decision making, aiming to bridge rigorous theoretical guarantees with practical deployment in high stakes settings.
A central part of my research investigates individual fairness in clustering, where similar individuals should receive similar outcomes. I develop algorithmic frameworks that incorporate fairness constraints into clustering objectives while maintaining computational tractability and solution quality.
My work includes linear programming based approximation algorithms for fair k-clustering with outliers, as well as extensions to correlation and hierarchical clustering under fairness constraints. These contributions provide provable guarantees while addressing challenges arising in real world datasets.
Real world datasets are often noisy, incomplete, and biased. I investigate how such imperfections affect learning algorithms and develop methods that remain robust under distributional irregularities. This includes handling outliers in fairness constrained optimization and designing algorithms resilient to structural noise.
I have worked on large scale public health datasets in collaboration with the Government of Gujarat, analyzing child malnutrition and maternal health data. This involved addressing noisy and incomplete data to support policy decisions in resource constrained environments.