U-Plan
AI-driven urban heat island mitigation — HackHarvard 2024 winner 🏆
1st Place, Sustainability Track — HackHarvard 2024. Selected from over 1,000 students across 284 universities and 22 countries.
U-Plan's interactive 3D GIS dashboard — urban heat, vegetation, and water coverage across Phoenix ZIP codes.
I led a team of four to build a city-scale sustainability platform analyzing urban heat, vegetation, and water coverage across 50+ Phoenix ZIP codes using Segment Anything Model (SAM) segmentation, GIS processing, and remote-sensing analytics.
- Built end-to-end Python pipelines and interactive 3D GIS visualizations
- Computed environmental indices (NDVI, NDWI, LST) from satellite data
- Generated localized heat-mitigation strategies using LLM-powered recommendations
SAM-based semantic segmentation for urban density understanding — classifying vegetation, water, and built surfaces from satellite imagery.
Stack: Python · Segment Anything Model · LLMs · Mapbox · Rasterio · Geopy · Folium