I am currently in my first year of my PhD in Computer Science at Unversity of Illinois Urbana-Champaign. My broad research interests lies in the area of improving the performance and reliability of distributed systems. I am currently working on improving the performance of resource managers and autoscalers for cloud-native microservice systems. I am a part of the Distributed Protocols Research Group (DPRG) and is fortunate to be advised by Prof. Indranil Gupta.
Previous to this, I have worked as a Research Associate 2 at Big Data Intelligence Lab, Adobe Research, India. I was a part of the 'Systems and Insights Group' where I primarily worked in the Systems area. My work at Adobe focused on varied projects. My main focus was towards maintaining reliability of cloud serices by applying machine learning and causal techniques, wherein some of my projects have been successfully transferred to the engineering team and is being used in the products. Apart from this, I have also worked on scheduling batched queries across different configuration of clusters and explored areas of using data for digital marketing experiences.
I graduated with a Dual Degree (Integrated B. Tech + M. Tech) in Computer Science and Engineering from the Indian Institute of Technology (IIT)Kharagpur. I was advised Prof. Sandip Chakraborty on my Master's Thesis on the topic of training machine learning models in a blockchain-based cross-silo federated network via blockchain interoperability and my bachelor's thesis project on the topic of viewport adaptive 360-degree video streaming.
I hail from Kolkata, West Bengal, India. Apart from my technical prowess, I am an avid reader (goodreads), sports enthusiast (especially football - P.S. a huge Barcelona and Messi fan), an adventure lover and a connoisseur of art and film-making. I like to read books and some of my favourite reads are 1984, To Kill a Mockingbird, Slaughterhouse-five and A Brief History of Time. In addition to that few of my favourite movies are The Shawshank Redemption, Parasite, Fight Club and The Godfather.
Topic: Improving the execution pipeline of LLM Agents through Program Caching
Research Areas:
LLM Agents, Caching, ML for Software Engineering, NLP
Topic: Architecting Asynchronous Federated Learning
Research Areas:
Federated Learning, Asynchronous FL algorithms, On-Client Training, Distributed Framework design
Topic: Advanced Optimization Methods for Machine Learning
Research Areas:
CANDECOMP/PARAFAC Decomposition, Sparse Tensor Completion, Leverage Scores, Randomized Algorithm
Topic: Mean Time to Failure and Early Detection of Equipments (Predictive Maintenance)
Research Areas:
Clustering Algorithms, EM Algorithm, Density Estimation, Gaussian Mixture Model
Several concepts regarding thread management and their scheduling
Inter and Intra process scheduling algorithms for SMP systems and their analysis
The following analysis is my interpretation of the ICLR 2017 paper “Understanding Deep Learning requires Re-Thinking Generalization”(arXiv link). This paper was awarded one of the three Best...