About Me

I am a Staff Research Scientist working on machine learning for large-scale production systems.

Before this role, I worked at LinkedIn, Netflix, and CognitiveScale as a Machine Learning Engineer and Research Scientist. I hold a Ph.D. in Electrical and Computer Engineering from The University of Texas at Austin, where I was advised by Prof. Joydeep Ghosh and Prof. Raymond J. Mooney. Before UT Austin, I spent four years on my undergraduate studies at Jadavpur University in Kolkata.

Research Interests

• Recommender systems and large-scale ranking
• Large-scale optimization and efficient deep learning
• Probabilistic and Bayesian machine learning
• Graph representation learning

Selected Work

1. Practical Design of Performant Recommender Systems using Large-scale Linear Programming-based Global Inference
   Casts global recommendation constraints as a large-scale linear program solved at production scale.
   KDD 2023
2. QuantEase: Optimization-based Quantization for Language Models
   A layer-wise, optimization-based scheme for post-training quantization of large language models.
   arXiv 2023 arXiv
3. A Precise Characterization of SGD Stability Using Loss Surface Geometry
   Characterizes the stability of stochastic gradient descent through the geometry of the loss surface.
   ICLR 2024 arXiv
4. A Dual Markov Chain Topic Model for Dynamic Environments
   A Bayesian topic model that tracks how topics and their prevalence drift over time.
   KDD 2018
5. Nonparametric Bayesian Factor Analysis for Dynamic Count Matrices
   A nonparametric Bayesian factor model for sequences of count matrices with an unbounded latent dimension.
   AISTATS 2015

See all publications →

Recent News

• Sep 2025 Paper on relational deep learning for transaction categorization accepted at ECML-PKDD 2025.
• Jan 2024 "A Precise Characterization of SGD Stability Using Loss Surface Geometry" accepted at ICLR 2024.