Shourya Bose

Logo

Homepage

View My Google Scholar Profile

Hi, I am Shourya Bose. I am a PhD candidate with the eODAL Lab at University of California, Santa Cruz. My advisor is Dr. Yu Zhang. My research interests span the intersection of machine learning and optimization applied to electric power systems. Within machine learning, my works deal with foundation models, reinforcement learning, and machine-learning aided optimization.

Previously I was a Research Associate (RA) at the Department of Electrical Engineering at Indian Institute of Science (IISc) Bengaluru in the Control and Network Systems Group.

I graduated with a Dual Degree from BITS Pilani (KK Birla Goa Campus) on December, 2019. I majored in M.Sc. (Hons.) Mathematics, and B.E. (Hons.) Electrical and Electronics Engineering. I graduated with a First Class grade in both the subjects.

You can download my CV here.

Jump to section: Domain Experience, Publications, Miscellaneous

Domain Experience

Machine Learning

As a part of my Ph.D. research and an internship at Argonne National Laboratory, I have carried out significant research on various aspects of machine learning applied to electric power systems. I have researched (and won an international competition) on the application of reinforcement learning for operation of power grids. I worked on the problem of load forecasting using time-series foundation models, and am currently concluding a project wherein I designed adapters for merging time-series foundation models with weather foundation models to create weather-informed load forecasters. Lastly, I worked on machine-learning accelerators for optimization problems which involved creating mixture-of-experts models that accelerate optimization problems by eliminating non-informative constraints, with each expert having a good understanding of different kinds of non-informative constraints.

Numerical Optimization

I have conducted research on different kinds optimization problems with applications for electric power systems. These works can broadly be divided into two categories - proposing new problem structures and solving existing problems more efficiently. With respect to the former, I have worked co-optimization of battery energy storage systems (BESS) with load restoration, and privacy-preserving load restoration using customer batteries. With respect to the former, please see the Machine Learning section for work on accelerating optimization problems using neural networks.

Formal Methods

I am currently enrolled in a graduate-level formal methods course. As a part of the same, I am working on two different projects:

Control Theory

I have worked on event-triggered control of linear systems over noisy channels between the sensor and the controller, resulting in a journal publication. I have also taken graduate courses on nonlinear and optimal control.

Journal Publications

Authors: Shourya Bose, Kejun Chen, Yu Zhang. Energy and AI, 2025

Authors: Kejun Chen, Shourya Bose, Yu Zhang. IEEE Transactions on Industrial Informatics, 2025

Federated Short-Term Load Forecasting with Personalization Layers for Heterogeneous Clients (arXiv)

Authors: Shourya Bose, Kibaek Kim. Report.

Authors: Shourya Bose, Yu Zhang. IEEE Transactions on Control of Network Systems, 2023. Received INFORMS ENRE Early Career Best Paper Award.

Authors: Shourya Bose, Pavankumar Tallapragada. IET Control Theory and Applications, 2021.

Conference Publications

Authors: Shourya Bose, Yijiang Li, Amy Van Sant, Yu Zhang, Kibaek Kim. NeurIPS Workshop on Time Series in the Age of Large Models, 2025

Authors: Shourya Bose, Yu Zhang, Kibaek Kim. IISE Annual Conference and Expo, 2024.

Authors: Shourya Bose, Yu Zhang, Kibaek Kim. IEEE PES-GM 2024.

Authors: Shourya Bose, Kejun Chen, Yu Zhang. IEEE ISGT 2023. Received IEEE SYPA Travel Grand, UCSC Dean’s Travel Grant.

Authors: Shourya Bose, Sifat Chowdhury, Yu Zhang, IEEE PES-GM 2022.

Authors: Shourya Bose, Yu Zhang, IEEE ISGT NA 2022.

Authors: Shourya Bose, Pavankumar Tallapragada. Fifth Indian Control Conference, 2019.

Competition Victories

INFORMS Data Challenge 2025 - First Position

Internation competition with over 40 teams with the objective of forecasting electrical outages for 83 counties in Michigan over the span of 48 hours. Our team (myself, Tomas Kaljevic, Yu Zhang) preented a solution based on a two-stage forecasting model powered by gradient boosted decision trees.

Learning to Run a Power Grid (L2RPN), 2023 - First Position

International competition with over 90 participants with the objective of using reinforcement learning based techniques to operate a simulated power grid for 24 hours. Our team (myself, Qiuling Yang, Yu Zhang) presented a solution which used a mixture of classical optimization, Monte Carlo search, and policy optimization.

Miscellaneous

Scholarships and Fellowships

As a part of my PhD journey, I am the grateful recipient of two competitive scholarships - the Chancellor’s Fellowship covering the first year of my research, and the Dissertation-Year Fellowship covering the final year of my studies.

Space

I am a huge space enthusiast! While I mostly follow space launches and space science in a private capacity, I had the opportunity to contribute to Pixxel Space as an intern in 2019. As a part of that internship, I understood some of the factors behind sizing and orbit choice for a constellation of earth observation satellites.