PhD student in Computing and Information Science at the Rochester Institute of Technology.
Advised by Prof. Xumin Liu and Prof. Qi Yu.
I currently work on knowledge distillation and large language models. My research interests lie in machine learning and artificial intelligence, especially computer vision, LLMs and knowledge distillation.
Before my PhD, I worked on handwriting recognition for low-resource scripts.
A young Sudip at a café in Kathmandu, circa 2026.
Started my PhD in Computing and Information Science at RIT, advised by Prof. Xumin Liu and Prof. Qi Yu.
Our paper on handwritten Devanagari conjunct consonants was accepted at the 17th IAPR International Workshop on Document Analysis Systems (DAS 2026).
Graduated with a BE in Computer Engineering from Pulchowk Campus. My thesis received the department's Best Final Year Project Award, chosen from 32 theses.
Accepted 17th IAPR International Workshop on Document Analysis Systems, 2026
A new benchmark of 50,020 labeled handwritten conjunct consonants, with transfer learning baselines that reach 99.57% on DHCD, above the previous 99%.
Information Systems for Intelligent Systems, Springer
A two-stage system for Nepali license plates. YOLOv8 extracts the plate and detects characters, then a CNN recognizes each one.
A GPT-2 model built from the ground up and trained on the Tiny Shakespeare dataset.
Semantic search over legal texts so people can ask legal questions in plain language. Winner of HackADev 2023, organized by UNDP, Citra Labs, Diyo.AI and NAAMII.
Random forest and logistic regression models that predict 30-day readmission for diabetic patients (AUC 0.7036), with a fairness evaluation across age, race and gender subgroups.
An LSTM-based next-word suggestion tool for Nepali, trained on a public corpus.
Transfer learning for skin lesion classification on the ISIC dataset.
A NumPy implementation on the Iris dataset, written as a guide to the mathematics and optimization behind SVMs.
2026 to present
PhD, Computing and Information Science
Rochester Institute of Technology, USA
Advised by Prof. Xumin Liu and Prof. Qi Yu.
2021 to 2025
BE, Computer Engineering
Pulchowk Campus, IOE, Tribhuvan University, Nepal
Graduated with 85.63%. Received the government merit scholarship for all four years, awarded for ranking 27th among 15,000 applicants.
2023 to 2024
Instructor
LOCUS and IT Club, Pulchowk Campus
Taught data science and machine learning fundamentals in the Software Fellowship, and CPU architecture and low-level C in the Advanced C course.
2022 to 2023
Machine learning intern
Alternative Technology, Kathmandu
Led a team of interns testing super-resolution algorithms to improve carpet image quality.
2020 to 2022
Executive head
Mathematics Initiatives in Nepal
Started Education To All, a program offering math, technical skills and career counseling at an orphanage. Organized the National Mathematics Olympiad, the selection round for Nepal's IMO team.
2025
Best Final Year Project Award
Selected from 32 theses across the department.
2023
HackADev winner
Presented and built Niti, the winning legal chatbot idea.
2023
Evidence Based Approach, Minamata
Represented Tribhuvan University in a 10-day research and fieldwork program hosted by Keio University, Japan.
2018 to 2021
Mahatma Gandhi Scholarship Scheme
Awarded by the Embassy of India for high school studies.
I'm happy to talk about research and collaborations, especially on knowledge distillation and LLMs.
st7013 at rit dot edu
Rochester, NY, USA
