About
I'm a graduate student in Computer Science at the University of Kansas, working where software engineering meets data and machine learning. My research and engineering sit at the intersection of the two — designing reliable systems that process, analyze, and learn from large, messy datasets.
At Smart Data Solutions, I engineered pipelines integrating Java and Python to process scanned medical documents, applying layout-aware transformers and LLMs to automate structured information extraction. Earlier, at Naamche, I built LLM- and retrieval-based systems combining semantic search and vector databases for real-time knowledge retrieval.
I care about scalable, data-driven systems that bridge infrastructure and intelligence — distributed systems, data engineering, and applied ML — and about making data processing faster, smarter, and more reliable.
News
- Aug 2025Began my MS in Computer Science at the University of Kansas as a Graduate Teaching Assistant.
- Jul 2025Wrapped up 1.5 years as a Machine Learning Engineer at Smart Data Solutions.
- Jan 2025Our paper on structured information extraction from Nepali scanned documents was presented at CHiPSAL @ COLING 2025, Abu Dhabi.
- 2024EaseAnnotate, a document-AI tool I helped build, launched as a SaaS and was backed by Microsoft for Startups.
Publications
COLING
2025
Structured Information Extraction from Nepali Scanned Documents using Layout Transformer and LLMs
Experience
Graduate Teaching Assistant
Lead weekly labs on C/C++, Git, and Docker for 30+ undergraduates; grade assignments, in-class problems, and labs.
Machine Learning Engineer
Built information-extraction systems for healthcare documents; shipped an intelligent medical-record processor using LayoutLM and LLMs, from data collection through model training.
Machine Learning / Backend Engineer
Built LLM chatbots and RAG systems over a vector database (Scrapy, Metabase, SageMaker); shipped FastAPI services with guardrails.
Machine Learning Intern
Worked on NLP with RNNs and transformer-based models, focusing on attention; built chatbots with the Rasa framework.
Technical Coordinator
Built and launched the online certification system, managed Azure infrastructure, and led a 10-day fellowship for 120+ students.
Education
MS, Computer Science
Analysis of Algorithms · Advanced Data Science · Static Analysis
BE, Computer Engineering
Artificial Intelligence · Data Science · Databases · OOP (C++) · Operating Systems · Distributed Systems
Selected Projects
EaseAnnotate ↗
Document-AI SaaS for annotation and key-information extraction. Backed by Microsoft for Startups.
Receipt Scanner ↗
LayoutLM-based key-information extraction from unstructured receipts, with CI/CD and Docker.
Anubadak ↗
Interpreter for a B-like language — lexing, semantic analysis, and AST construction.
3D City Modeling ↗
A 3D renderer from scratch in C++/OpenGL with Gouraud and Phong shading.
Real-time Text Similarity ↗
Sentence-embedding similarity search (TF-IDF, SIF, BERT) with Annoy for approximate nearest neighbours.
Sorting Visualizer ↗
An interactive visualizer for sorting algorithms, built in C++.
Writing
Loading posts…
Teaching
Small, visual tutorials that teach the fundamentals from the ground up. I started them for my younger brother, and I keep making them because I genuinely love it. They live on neustack.net. Each was built with AI as a hands-on collaborator — from drafting the explanations to generating the interactive visualizations.
Learning Python in the Age of AI
Learn Python the modern way — writing real code, using AI as a tool, and building intuition that lasts.
Math, the Visual Way
From derivatives to complex numbers — no scary formulas, just pictures, sliders, and intuition.
Physics from Vectors Up
A visual course that starts with vectors and builds intuition for motion, force, and fields.