Myaka Akhila

CNN-based web app that identifies plant diseases from leaf images with real-time inference — helping farmers detect crop issues instantly.

• Trained a CNN model across 15+ disease categories achieving ~90% classification accuracy on validation data
• Built a Flask backend with image upload, preprocessing pipeline, and model inference — <2s end-to-end response time
• Delivered real-time predictions through an integrated HTML/CSS frontend with zero page-reload inference

ML-powered Flask web app that predicts kidney stone risk from patient medical data — enabling faster screening for clinical use.

• Trained a binary classifier on 10+ patient health parameters achieving ~87% prediction accuracy with high recall
• Engineered a REST-style request-response workflow — form submission to result rendered in under 1 second
• Server-side validation and preprocessing in Flask eliminated 100% of malformed inputs before model inference

JavaFX desktop app for vocabulary building — combines word tracking, progress logging, and a spaced-repetition quiz system in a single modular application.

• Designed a complete OOP architecture with 4+ decoupled modules (storage, UI, quiz logic, word management)
• Implemented 100% persistent data retention across sessions using core Java file I/O — zero data loss on restart
• Spaced-repetition quiz engine boosts retention by up to 40% vs passive review, with per-word progress tracking