Lalit Kishore Vyas

Based in Mumbai

I am a Lead Data Scientist at Motilal Oswal Financial Services in Mumbai, India, where I work on building recommender systems and personalization solutions.

My research interests lie at the intersection of Machine Learning, Information Retrieval, and Recommender Systems, with a specific focus on user modeling and personalization. I am passionate about developing intelligent systems that understand user preferences and deliver tailored experiences.

My current research interests focus on fairness-aware learning in graph neural networks, developing regularization frameworks that mitigate bias in learned node representations across sensitive attributes.

I am pursuing a Masters from Liverpool John Moores University, UK, and hold a B.Tech from Pandit Deendayal Energy University (formerly PDPU), India.

Selected Publications

UMAP

Addressing Personalized Diversity in Eyewear Recommendation: a Lenskart Case Study

Lalit Kishore Vyas and Ludovico Boratto

In Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization, New York, USA, 2025

Abs DOI

This study addresses the challenge of limited diversity in recommender systems on e-commerce category pages, which often leads to reduced user engagement and satisfaction. Recognizing the limitations of traditional Factorization Machines (FM) in generating diverse recommendations, we propose a personalized diversity approach that combines re-ranking strategies with FM, enhanced by Generalist-Specialist (GS) scores to tailor diversity to individual user preferences. The re-ranking strategies explored include Maximal Marginal Relevance (MMR) and Determinantal Point Processes (DPP). Our results show improved balance between relevance and personalized diversity in offline experiments. Additionally, we investigate an alternative approach to personalized diversity through a contextual bandit model (LinUCB), where diversity emerges by balancing exploration and exploitation in predicted preferences. This evaluation highlights LinUCB’s ability to anticipate diverse recommendations by simulating adaptive responses without relying on active user feedback, offering a contrast to traditional re-ranking methods.