AI Enginner (Intern)

Highlights

• Modified and optimized a YOLO-based model to accept >3 channel TIFF satellite imagery (up to 13-channel Sentinel-2 data), conducting rigorous channel combination experiments to maximize the segmentation accuracy of solar panels across diverse geographic locations.
• Engineered and deployed an integrated web-based pipeline utilizing Google Maps and Google Earth Engine (GEE) for automated satellite data retrieval and model inference, providing users with instant solar panel segmentation and supplementary weather condition reports for solar optimization forecasting.