Jaswanthreddy gongati

At Codec Technologies, I built a personalized film recommendation system using content-based and collaborative filtering. I combined matrix factorization with cosine and Pearson similarity metrics to address cold-start problems.

I preprocessed movie data, extracted features, and evaluated recommendation accuracy. I also deployed the engine through an interactive web interface for real-time user testing.

On a project, I designed an end-to-end pipeline that predicts cardiovascular risk from retinal fundus images. The Dual Vessel-Aware Transformer Network combines local vessel patterns with global image context.

I benchmarked the model with AUC-ROC and F1-score, with performance exceeding baseline CNN architectures. I also used Grad-CAM and attention maps to generate visual explanations for clinician validation.