Anju Chhetri

About

Hello, my name is Anju Chhetri.

I am an engineering graduate and currently a research assistant at Multimodal Learning Lab at NAAMII. My current research interests are in building more robust and reliable machine learning models. In the past, I have worked in the domain of adversarial machine learning, active learning and computer vision. In general, my passion centers around understanding the inner workings of machine learning systems.

When I'm not working on research, I enjoy being surrounded by nature, often going for hikes. Apart from hiking, I find peace in painting and playing my ukulele.

Work Experience And Research

Research Assistant

NepAl Applied Mathematics and Informatics Institute for research (NAAMII)

Advised by Dr. Binod Bhattarai

(April, 2024 - Present)

Working on research projects focused on machine learning robustness and interpretable AI.
Developed an OOD detection method for pathology segmentation in volumetric medical images, specifically MRI and CT scans, based on self-supervised learning with synthetic anomalies, leading to a 5th-rank in the MICCAI 2024 challenge.

Undergraduate Research Intern

NepAl Applied Mathematics and Informatics Institute for research (NAAMII)

Advised by Dr. Binod Bhattarai

(August, 2023 - March, 2024)

Worked on an Active Learning framework to determine the most effective uncertainty-based sampling technique to address dataset imbalance in a multi-class object detection task.
Trained deep learning models and conducted an ablation study for detecting gait abnormalities in ware- house environments.

Projects

Improving Adversarial Image Robustness For Preventing Image Manipulation

(October 2023 – April 2024)
- Developed an effective JPEG-resistant image immunization pipeline to counter text-guided AI manipulation, specifically for diffusion models. Additionally, studied the influence of text prompts on the success of image immunization against these text-guided diffusion models.
- Code
Real-time View-Space Grass Generation for Video Games using GAN

(December 2022 – April 2023)
- My team and I developed a real-time neural rendering technique for generating realistic grass in video games. My task was to experiment with existing architectures, analyze results to identify optimization opportunities, and design new architectures that achieved both visual fidelity and low inference time.
- Paper | Code

Scholarships and Awards

Award/Scholarship	Awarded By	Year
Mentee, Women Leaders Mentorship Programme (4.4% acceptance rate)	EU in Nepal, Mentor: H.E. Veronique Lorenzo, EU Ambassador to Nepal	December 2024–Present
5th Place, Pixel-level Task, Medical Out-of-Distribution Analysis Challenge	MICCAI 2024	October 2024
2nd Runner Up - Hacking for Humanity 2023	Shequal Foundation	Februrary 2023
Winner - Gritfeat AI Health Hackathon	Gritfeat	January 2023
Winner - Ace Ignite	Ace Institute of Management	November 2022
2nd Runner Up - ABU Robocon	ABU Asia-Pacific Robot Contest (ABU Robocon)	August 2022
Selected Participant - GitHub Field Day Nepal	GitHub Education	June 2022
Full scholarship recipient of NAAMII Winter School of AI, 2021	NepAl Applied Mathematics and Informatics Institute	December 2021
Mahatma Gandhi Scholarship	Embassy of India	2017-2018
Awarded Paramount Gold Medal	Paramount Boarding Higher Secondary School	2018

Anju Chhetri | अन्जु क्षेत्री

About

Work Experience And Research

Projects

Scholarships and Awards