Built and deployed a full-stack AI image analysis platform with Google Gemini, supporting image captioning, detailed analysis, OCR, object detection, scene analysis, and vision chat. Added JWT authentication, private analysis history, responsive tabs, CORS, environment-based secrets, and production deployment across Vercel and Render.

Developed a deep learning model using the UNet architecture to segment brain tumors from MRI images. Achieved a Jaccard index of 0.73 and Dice score of 0.82, demonstrating strong accuracy in tumor boundary detection through regularization and data augmentation techniques.