• Developed a custom WordPress theme with animations using Vanilla JavaScript and CSS.
• Integrated Google Maps API to display multiple location pins.

• Utilized the PVIT model to classify impacted teeth in proximity to the IAN nerve, achieving 95% accuracy
(compared to ResNet’s 89%).
• Applied Dino V2 for precise segmentation of the IAN nerve and impacted regions.
• Fine-tuned pre-trained models for enhanced detection performance on dental imaging datasets.

• Developed a web-based diagnostic system for detecting mental health conditions (depression, suicide, stress)
using machine learning models (AdaBoost, SGD, Naive Bayes) achieving over 96% accuracy.
• Integrated models using Streamlit and deployed a basic NLP chatbot to provide mental health tips and
guidance.

• Processed and annotated a custom cataract dataset with categories for various cataract types (normal, cortical,
nuclear, immature, mature) using Roboflow.
• Leveraged instance segmentation models (Yolov11, SAM2, Mask R-CNN) to classify cataract types and determine
the infected area as a percentage of the total region.
• Designed a pipeline for classification and segmentation, integrating novel dataset preparation techniques to
improve detection accuracy and performance.