I am currently pursuing an Integrated MS at IISER Thiruvananthapuram, and this post summarizes the research agenda that guides my present work.
Core Questions I Am Working On
- How can physics-informed graph models improve reliability in geospatial risk prediction?
- How can remote sensing and ground-sensor streams be integrated for operational decision support?
- How can we design data products that remain scientifically rigorous while being deployable in public systems?
Ongoing Research Themes
Physics-Informed Graph Learning for Fire-Risk Analysis
I have been developing PI-GNN based workflows to model spatio-temporal wildfire spread in high-risk regions of Kerala. My objective is to bridge physical priors and data-driven inference so that predictions remain interpretable under changing field conditions.
Time-Series Forecasting for Monsoon Onset
Using observational datasets, I have worked on improving monsoon onset prediction for Central India through structured time-series modeling. This line of work focuses on practical forecast accuracy and uncertainty-aware interpretation.
Synthetic Terrain and Geospatial Simulation
I have also explored reproducible terrain generation methods for geospatial analytics and simulation. This helps stress-test models in controlled yet realistic settings before field deployment.
Research Orientation
I approach research as an interdisciplinary exercise involving machine learning, domain knowledge, and implementation constraints. My long-term focus is to build systems that are technically strong and societally useful.