I'm an undergraduate pursuing a B.Tech in Artificial Intelligence at Kathmandu University, driven by a deep interest in using AI to solve real-world problems, particularly in low-resource settings like Nepal. My work spans deep learning, computer vision, and multilingual NLP, with hands-on research experience ranging from fine-tuning vision-language models for scientific figure understanding to building multilingual news summarization systems and deep reinforcement learning agents for traditional board games. I'm especially motivated by problems where limited data or resources demand creative, efficient solutions, whether that's achieving state-of-the-art handwriting recognition with a fraction of the usual parameters or deploying lightweight models for on-device inference.
Beyond research, I enjoy building end-to-end applications that bring AI into practical, everyday use, from face recognition tools for photographers to mobile apps that classify Nepali cuisine in real time. I've also taken on coordination roles within Kathmandu University's AI Club, organizing technical events and workshops for 100+ participants. Whether working independently on research or collaborating across teams, I care about problem-solving, clear communication, and building AI systems that are both technically rigorous and genuinely useful.
Feel free to connect.
Projects
Scarlet
Lightweight Qwen2-VL fine-tuning pipeline optimized for low-memory GPUs using QLoRA, LoRA, and 4-bit quantization.
Nepali News Aggregator & Summarizer
Real-time news aggregation platform that fetches, classifies, and displays news from multiple sources.
BaghChal
Reinforcement learning agents for the traditional Nepali board game BaghChal.
REACO
AI-powered app that detects, classifies, and organizes photos by individuals using face recognition.
Parikar
Mobile app that identifies 12 Nepali dishes using AI-based image classification.
Publications
Side Quest
FieldSeg
AI-powered system for detecting and segmenting agricultural field boundaries from remote sensing imagery.
Kalam
Simple yet powerful Nepali text editor bringing native language writing to the modern web.
Multimodal RAG
Multimodal RAG assistant that retrieves images using CLIP embeddings and answers with LLaMA 3.
Scholar Summary
AI research paper summarizer that uses LLaMA 3 to generate customizable academic summaries.
Chat Bot
AI chatbot using LangChain and LLaMA 3 for intelligent conversations.
Analysis of Meteorites
Analyzes meteorite fall patterns and trends using historical impact data.
Learn to CODE
Programming practice showcasing C, C++, Python, and ML projects from fundamentals to advanced.
DataSets
SCARLET
Scientific figure reasoning dataset with 29,887 figures and 157,387 multimodal QA pairs from 3,058 arXiv papers for Vision-Language Models.
Parikar (Nepali Food)
Nepali food image dataset with 15,250 images across 20 classes for food recognition tasks.
NEPSE Listed Companies
Nepal Stock Exchange dataset containing financial details of 520 listed companies for market analysis.
Competitions
Education
B.Tech in Artificial Intelligence
Kathmandu University, Dhulikhel, Nepal
+2 Science
Lord Buddha Secondary School, Biratnagar, Nepal
Secondary Education (SEE)
Elixir Academy, Kathmandu, Nepal