Background and direction
About
I am a researcher working at the intersection of computer vision, deep metric learning, and applied machine learning, with a focus on non-invasive biometric identification, legal information retrieval for low-resource languages, and image forgery detection.
Biography
I am a researcher specialising in computer vision, deep metric learning, and applied machine learning, with a particular focus on problems that require models to generalise across real-world, uncontrolled conditions.
My current research investigates non-invasive livestock biometric identification through Siamese neural networks, conducted at Plantsat in collaboration with the United Nations Development Programme (UNDP) in Nepal. I have designed and evaluated multiple backbone architectures — VGG16, ResNet50, ResNet152, ViT-Base, and ViT-Large — and applied Grad-CAM interpretability to validate that model attention aligns with anatomically meaningful muzzle features.
I have co-authored conference papers at ICAIL 2025 and ICAIL 2026 on AI-assisted legal information retrieval for Nepali legal documents, and conducted an independent comparative study of CNN and Transformer architectures for deepfake and image forgery detection, presented at NCCI 2025.
Research direction
Research interests
- Computer Vision & Deep Metric Learning
- Non-Invasive Biometric Identification
- Explainable AI & Model Interpretability
- AI & Law · Legal Information Retrieval
- Image Forgery Detection & Deepfake Classification
- Low-Resource Language NLP
Methods
- Siamese neural network modeling
- Contrastive loss optimization
- Explainable AI (Grad-CAM visualization)
- Multi-environment field data collection