LeafCure is an end-to-end AI-powered plant disease detection platform designed to help farmers, researchers, and agriculture enthusiasts identify plant diseases from leaf images. The application leverages Vision Transformer (ViT) models to classify diseases across 38+ categories, providing fast and reliable predictions through an intuitive web interface.

The platform was built with a focus on production-ready deployment rather than just model training. I optimized model loading, inference latency, and request handling to ensure a smooth user experience while serving predictions efficiently. The backend exposes REST APIs using FastAPI, making it easy to integrate with other applications or mobile clients.

To improve deployment performance, I profiled model execution, optimized inference workflows, and implemented efficient preprocessing pipelines for image normalization and transformation. The result is a scalable AI application that demonstrates how modern transformer-based architectures can be deployed for real-world agricultural use cases.