Hi, thank you for visiting my home page!

I’m a PhD student in Computer Science at the University of Illinois at Chicago (UIC), under the advisement of Professor Aadirupa Saha, where I explore the theoretical foundations and practical applications of intelligent decision-making systems. My research lies at the intersection of reinforcement learning, convex optimization, and machine learning theory, with a particular focus on how agents can learn efficiently in continuous, complex environments.

Research Interests

My work centers on building principled, computationally-aware learning systems that bridge theory and practice:

• Abstraction in Reinforcement Learning: Developing hierarchical frameworks that enable agents to reason at multiple levels of granularity, reducing computational complexity while maintaining performance guarantees

• Convex Optimization for Neural Networks: Investigating optimization landscapes and convergence properties of deep learning systems through the lens of convex analysis

• Learning in Continuous Environments: Designing algorithms for sequential decision-making in high-dimensional, continuous state and action spaces

• Multi-Agent Learning & Communication: Exploring coordination mechanisms and emergent behaviors in systems where multiple intelligent agents interact

Current Work

I’m currently working on combinatorial multi-armed bandits with monotonic aggregation functions. Traditional bandit algorithms assume linear reward structures, but real-world decision-making, from personalized recommendations to healthcare treatment selection, that often involve complex, non-linear aggregations.

My research introduces novel algorithms that efficiently identify optimal arm subsets without requiring knowledge of the underlying aggregation operator. By cleverly reducing combinatorial feedback to pairwise comparisons through randomization and selective sampling, we achieve sample complexity matching state-of-the-art methods while eliminating many restrictive assumptions. This work bridges pure exploration theory with practical applications where ranking and aggregation matter as much as selection.

Recent Publication: Efficient Algorithms for Combinatorial-Bandits with Monotonicity (OPT 2025, co-authored with Aadirupa Saha)

Previously, I’ve explored:

• Value smoothing using similarity-based latent embeddings for reinforcement learning

• Metareasoning and computational efficiency in Monte Carlo Tree Search

• Applications of RL to bug detection in game environments

Background

Before joining UIC’s PhD program (originally as an MS student, later transferred), I earned my undergraduate degree from Manipal University Jaipur (2015-2019), where I developed foundational skills in machine learning and systems. I’ve also worked in various machine learning roles in the industry before choosing to focus on the research aspects of ML.

What Drives Me

I believe the future of AI lies in systems that are not just powerful, but interpretable, efficient, and theoretically grounded. Whether it’s understanding the partially convex structure underlying neural network training or designing RL agents that plan with limited compute, I’m passionate about research that connects elegant mathematical theory with real-world impact.

When I’m not figuring out math or reading papers, you’ll find me playing the piano, cooking, or running.