Mihir Bulsara

Architected CNN-based deep learning pipeline for urbanization detection from Sentinel-2 multispectral imagery achieving 85%+ accuracy across 5+ Indian metropolitan regions for Spatio-Temporal Building Probability Mapping project; Engineered automated pre-processing workflows with OpenCV for pixel-level geometric alignment and deployed on Google Cloud Platform handling 500GB+ satellite datasets; Developed full-stack Vue.js and Python modules for Fasal 2.0 and Krishi Decision Support System Dashboard serving 15+ Indian states with real-time crop analytics for national agricultural monitoring; Optimized PostgreSQL database queries reducing page load times by 40% for 100+ concurrent users and implemented interactive Chart.js visualizations with drill-down capabilities; Built real-time Bihar Flood Alert Dashboard integrating Sentinel-1 SAR data with automated threshold alerts and geospatial heatmaps covering 40+ districts

Enhanced climatology datasets by reconstructing missing coastal pixels using machine learning algorithms, Ferret visualization software, and GrADS data analysis tools; Processed 10+ years of NetCDF satellite data improving data quality and coverage for climate analysis and environmental monitoring applications