Building End-to-End ML Pipelines
A deep dive into creating robust, automated machine learning workflows from data ingestion to model deployment using tools like Airflow.
Read More
A passionate Computer Engineering student interested in AI, Data Engineering. I am learning to build scalable, AI-powered solutions to solve real-world problems.
I am a Computer Engineering student at Kathmandu University, with an exchange semester at IIT Palakkad. My passion lies in creating intelligent systems that make a real-world impact.
As a Co-Founder and leader, I've developed AI-powered solutions like Dr.Fish, gaining invaluable experience in building scalable systems and delivering impactful technical solutions. My work has been recognized with prestigious awards including the NASA Space Apps People's Choice.
I'm driven by curiosity and the desire to push the boundaries of what's possible with technology, always focusing on solutions that benefit society.
Kathmandu University & IIT Palakkad
AI, Data Engineering
Hult Prize Finalist
Built a parallel corpus (≈10k sentence pairs) from Nepali and English legal texts using OCR (PyTesseract) and manual alignment/verification performed with law-student collaborators.Fine-tuned MBART50 and custom Transformer variants for domain adaptation; baseline BLEU around 0.42 (ongoing improvements focused on formatting, legal phrase preservation, and domain-specific tokenization)
Assembled a mixed dataset (manipulated + AI-generated) totaling over 140k images across multiple manipulation types for benchmarking image-forensics approaches. Implemented and benchmarked convolutional CNNs (InceptionV3 family) vs Vision Transformers; re- ported improved accuracy and efficiency for selected CNN baselines in our experiments (InceptionV3: 94% on our test split)
Created synthetic ethical-dilemma datasets (≈1k cases) covering utilitarian, deontological, and virtue ethics perspectives for probing small LLMs. Fine-tuned TinyLlama-1.1B using LoRA-based methods to create specialized agents and evaluated consistency and robustness via human & automated metrics.
Proposed AlgoRAG, a specialized Retrieval-Augmented Generation system for theoretical computer science education, combining LLM understanding with domain-specific knowledge retrieval. Developed a knowledge base of textbooks, lecture slides, practice problems, proof examples, and complexity worksheets. Implements domain-specific optimizations for mathematical entity recognition, similarity metrics, pedagogical ranking, and proof query handling. Evaluated on 450 exam questions covering asymptotic analysis, recursive algorithms, dynamic programming, graph algorithms, and NP-completeness. Expected >85% accuracy and improved conceptual understanding compared to textbook-based learning.
Developed DerivativeAnimator, a fine-tuned LLM system to automatically generate Manim-based animations for calculus derivative concepts. Addresses accessibility challenges in STEM education by enabling educators without programming expertise to create high-quality visualizations. Collected 537 Manim code snippets, organized by complexity levels. Evaluated on technical correctness, pedagogical effectiveness, and educational impact. Improves student conceptual understanding by 20-35% and reduces animation creation time from hours to minutes, promoting educational equity in calculus learning.
Revolutionizing Aquaculture with AI
An AI+IoT startup focused on predicting algae blooms and fish diseases in aquaculture. Winner of Hult Prize Campus Competition and represented Nepal at the Global Summit in Bangkok.
AI-Driven Solutions in Mental Health & Sustainability
Currently developing two innovative AI-powered startups in the mental health and sustainability sectors. Solutions leverage machine learning, NLP, and behavioral analytics to address critical societal challenges.
Deploying YOLOv8 object detection pipeline with Docker containerization. Researching NVIDIA DeepStream, GStreamer, and Triton Inference Server for real-time multi-camera frame processing.
Volunteering to develop AI-driven recommendation systems and backend infrastructure for a global mental health platform, focusing on scalable solutions for mental wellness support.
Successfully represented Nepal at the Hult Prize Summit in Bangkok, Thailand.
Contributed to national pension platform serving 600k+ users. Implemented Nepal's first governmental liveliness detection using Amazon Rekognition. Developed 60+ page Angular UI with Spring Boot backend.
Building custom transformers and fine-tuning large language models (LLMs) for translating Nepali legal documents into English, contributing to legal tech accessibility.
Pursuing advanced coursework in AI, Compiler Design, and Graph Theory. Current GPA: 7.61/10, focusing on cutting-edge research and academic excellence.
Pursuing Computer Engineering, focusing on AI, networking, and software development with hands-on project experience.
micro1
2025
Validated expertise in core principles, tools, and best practices for building modern AI and data-driven systems.
Nvidia & DataCamp
2025
Completed "Fundamentals of Deep Learning" (Nvidia) and "Deep Learning with PyTorch" (DataCamp) for hands-on neural network experience.
OpenHPI
2025
Gained specialized knowledge in Natural Language Processing, focusing on creating and utilizing vector representations of text.
DataCamp & Udemy
2025
Completed comprehensive courses covering Supervised & Unsupervised Learning with Scikit-learn and ML fundamentals.
Astronomer
2024
Certified in both Airflow Fundamentals and DAG Authoring, demonstrating expertise in complex data pipelines.
University of Helsinki & DataCamp
2024 - 2025
Completed "Elements of AI" for comprehensive AI overview and "AI Fundamentals & Data Literacy" certification.
Kathmandu University
Represented Nepal at the Hult Prize Global Summit in Bangkok for the AI-powered aquaculture startup, Dr.Fish.
People's Choice Award 2024
Awarded People's Choice for 'Cosmira,' an interactive 2D educational web app for science outreach and collaboration.
Biggest Tech Award of Nepal
Recognized as a finalist for significant contributions and innovative projects in Nepal's national technology ecosystem.
Health Track
Achieved first place for developing 'MastiskaTrack,' an innovative AI solution for mental health assessment.
Embassy of India, Nepal
Received prestigious scholarship for outstanding academic excellence and performance during high school studies.
Mathematical Competition
Reached semi-finals in Nepal's first Integration Bee, demonstrating strong mathematical and problem-solving abilities.
NASA Space Apps
Hult Prize
Kantipur Media
TechPana
Kathmandu University
A deep dive into creating robust, automated machine learning workflows from data ingestion to model deployment using tools like Airflow.
Read More
A practical guide on adapting large language models like MBART50 for specific tasks, based on my legal NLP research experience.
Read More
Exploring how tools like Docker, Terraform, and CI/CD can streamline the entire machine learning lifecycle and deployment process.
Read MoreI'm always open to discussing new opportunities, interesting projects, or just chatting about technology. Whether you have a project in mind, want to collaborate, or need consultation, feel free to reach out!
Dhulikhel, Kavre, Nepal
+977 9810538507
English, Nepali, Hindi