All Projects
Every project I've built — from full-stack apps to deep learning experiments.
Designed and fine-tuned a Swin Transformer deep learning model for accurate classification of skin diseases. Built pre-processing pipelines and validated model performance through test cases. Worked on software diagrams for proper end-to-end implementation and debugged to improve model accuracy.
Implemented and compared multiple architectures including U-Net Lite, U-Net Full, U-Net++, DeepLabV3+ (ResNet-50), HRNet-Small, and SegFormer-B0. Built a custom data preprocessing pipeline including resizing, normalization, binary mask generation, and data augmentation using Albumentations. Optimized training for limited GPU memory using mixed-precision training and efficient batch sizing.
A full-stack scholarship management platform built with Next.js and TypeScript. Features include AI-driven PDF parsing for bulk scholarship uploads, document management for student profiles, and a smart matching algorithm. Developed as a collaborative project with a team.
A TypeScript-based full-stack web application designed for health and fitness monitoring. Built with a modern frontend stack and focused on user-friendly UI/UX for tracking health metrics and fitness routines.
Built a machine learning system to predict injury risk in athletes using health and performance data. Developed as a 4th semester AI course project, featuring data analysis, model training, and risk scoring with Python and Jupyter Notebooks.
A JavaScript web application built for managing and hosting the CaseQuest 2025 case study competition. Features include team registration, case submission, and judging workflows.
Built an automated web scraping tool using Python and Selenium for extracting structured data from websites. Demonstrates browser automation, data parsing, and handling dynamic content.
A collection of computer vision mini-projects using Python and OpenCV, including real-time object detection, image processing, and feature extraction experiments.
A JavaScript frontend component implementing a search bar with lazy loading functionality for efficient data fetching and rendering. Demonstrates performance optimization techniques for large datasets.