Core Technologies
Experience
Software Developer Intern
August 2025 – Present@ Clonefutura
- Designing and developing a Check-in/Check-out web application at Online Live Learning (OLL), currently in production and used to track 50+ educators.
Software Developer Intern
January 2025 – June 2025@ Iron Willed Tech Ltd
- Built responsive UI/UX with Kotlin and Jetpack Compose
- Integrated Firebase APIs for authentication and dynamic content
- Implemented Google MediaPipe ML model for real-time face mesh generation
Research Intern
February 2024 - July 2024@ Indian Institute of Technology (IIT) Jammu
- Optimized YOLO models achieving 80.95% size reduction with minimal accuracy loss
- Developed PPE detection system with 83.75% parameter reduction
- Published research findings on model compression techniques
Bachelor of Technology in Computer Science
2021 - 2025@ University Teaching Department - CSVTU
- Specialized in Machine Learning and Computer Vision
- Developed multiple award-winning projects including plant disease detection app
- Active participant in coding competitions and hackathons
Scalable Project Gallery
Text Behind Photo
Text Behind Photo is a dynamic web application that allows users to creatively integrate text within their images.Built with a modern frontend stack of Vite, React, TypeScript, and Tailwind CSS, the application provides a seamless user experience for uploading photos.The powerful FastAPI backend then processes the image, utilizing a YOLO model for person detection and a SAM2 model for precise segmentation through the Ultralytics framework.After automatically separating the subject from the background, the application returns the distinct layers to the user.From there, users can add multiple text elements, each individually customizable, and cleverly place them either in front of or behind the main subject to create visually striking compositions.
Real-time Book Reading Attention Monitoring System
This project is a real-time attention monitoring system for readers. Using a webcam, it employs a YOLO model to detect the user's face and the book they are reading, while the L2CS-Net model analyzes their gaze to determine if they are paying attention.The system, built entirely in Python, can generate session reports with attention metrics and can be configured for different monitoring durations.For optimal performance, a CUDA- enabled GPU is recommended.
Model Comparison and Compression
Optimized YOLO models for PPE detection, achieving 80.95% size reduction while maintaining high accuracy. Highlights: Analyzed 6 YOLO object detection models for PPE detection on the CHV Dataset. Optimized YOLOv8n models, achieving an 80.95% reduction in model size and an 83.75% reduction in parameters, with minimal drop in precision and recall compared with YOLOv8n. Executed rigorous testing protocols to evaluate YOLOv8 effectiveness against real-world ppe datasets, identifying critical areas for improvement which led to more accurate PPE identification within operational environments.
Plant Disease Recognition Using Machine Learning
AI-powered mobile app for plant disease detection with 98.45% accuracy and integrated chatbot for farmer support. Highlights: Fine-tuned a pre-trained MobileNetV3 CNN model, achieving 98.45% training accuracy and 97.69% validation accuracy for plant disease recognition. Engineered an intuitive Android application leveraging Flutter, designed specifically for farmers; integrated a plant-specific chatbot, resulting in over 100 unique queries answered during testing.