YashKakadiya

Problem: Farmers often misdiagnose plant diseases, leading to crop loss and overuse of chemical pesticides.

Solution: Trained a custom CNN and utilized ResNet50 Transfer Learning on a dataset of 54K+ augmented images to automatically classify diseases from leaf scans.

Impact: Achieved ~99.5% validation accuracy, providing an instant and reliable diagnostic tool.

Problem: Traditional artisans lack the technical skills to create engaging marketing content for the digital economy.

Solution: Built a Gen AI Platform using the Google Gemini API to automatically generate localized social media posts and product descriptions.

Impact: Empowered non-technical users to generate professional marketing copy instantly with zero learning curve.

Problem: Black-box medical models are often distrusted by healthcare professionals due to lack of transparency.

Solution: Developed a Voting Ensemble predictor (Logistic Regression, Random Forest, XGBoost) and integrated SHAP values to explain the impact of 12 clinical features.

Impact: Delivered an interpretable model that not only predicts risk but explains exactly *why*, aiding clinical decision-making.

Problem: Educational institutions struggle to track and predict student performance trends effectively across semesters.

Solution: Engineered a robust RESTful API with ASP.NET Core and a React dashboard featuring secure JWT authentication for role-based access.

Impact: Streamlined performance tracking, enabling educators to identify struggling students earlier in the semester.

Problem: Hospitals often rely on fragmented, legacy systems that slow down patient intake and staff coordination.

Solution: Architected a comprehensive HMS using ASP.NET Core MVC with real-time interactive dashboards powered by SignalR.

Impact: Centralized hospital operations into a single, scalable web application, improving administrative efficiency.