About
I am a research assistant at the NepAl Applied Mathematics and Informatics Institute for Research (NAAMII), working within the Multimodal Learning Lab under the supervision of Dr. Binod Bhattarai (Assistant Professor, University of Aberdeen; Co-founder, NAAMII; Chief Scientist, Fogsphere, UK). I hold an undergraduate degree in Computer Engineering from Tribhuvan University, IOE, Pulchowk Campus.
My research focuses on federated and multimodal learning for healthcare — developing privacy-preserving frameworks that enable collaborative model training across distributed clinical institutions. I am particularly interested in problems involving missing modalities, data heterogeneity, and robust multimodal fusion in low-resource medical settings. I'm also keen to bring machine learning to the edge.
I'm also involved in industry as a Deep Learning Engineer at Fogsphere (UK), where I design and train models for safety-critical environments.
Beyond research, I enjoy building things from first principles — from implementing cryptographic algorithms in C and writing a grep clone in Rust, to exploring CTF challenges and blockchain-based systems. I find the algorithmic side of computing deeply rewarding.
Multimodal Learning Lab — Supervised by Dr. Binod Bhattarai
Lalitpur, Kathmandu, Nepal
United Kingdom
News
- 2026 Whitepaper on PriFed released by Health AI for All Network (HAINet).
- 2026 Preprint released: Med-MMFL: A Multimodal Federated Learning Benchmark in Healthcare.
- 2025 Joined Fogsphere as Deep Learning Engineer (Part-time).
- 2025 Research Assistant at NAAMII, Multimodal Learning Lab.
- 2025 Paper accepted at MIUA 2025 (LNCS, Springer): Multimodal Federated Learning With Missing Modalities through Feature Imputation Network.
- 2025 Completed B.E. in Computer Engineering from IOE, Pulchowk Campus, Tribhuvan University.
- 2024 Joined NAAMII as Research Intern at the Multimodal Learning Lab.
Research Interests
My research centres on building federated learning systems that allow distributed clinical institutions to collaboratively train models without sharing sensitive patient data. I work on multimodal learning problems — particularly handling missing modalities and heterogeneous data distributions — with direct applications to medical image analysis and clinical decision support.
Publications
-
Med-MMFL: A Multimodal Federated Learning Benchmark in HealthcarePreprint — 2026
-
PriFed: Privacy-Preserved Learning in Health AI Through Federated LearningWhitepaper — Health AI for All Network (HAINet), 2026
-
Multimodal Federated Learning With Missing Modalities through Feature Imputation NetworkMedical Image Understanding and Analysis (MIUA 2025) — LNCS vol. 15916, pp. 289-299
Projects
Feature imputation network for multimodal federated learning with missing modalities — official codebase for our MIUA 2025 paper.
A basic implementation of the Unix grep utility in Rust — pattern matching, file searching, and command-line argument parsing.
Proof-of-concept decentralized voting system built on blockchain fundamentals — block hashing, chain validation, and vote integrity.
An alien language translation game built in Unity — shader-based visual effects and procedural dialogue generation.
Intentionally vulnerable web app demonstrating security flaws in language switching — built for CTF and education; demoed at InnoVision@NIT-Rourkela.
See more on GitHub →
Writing & CTF Writeups
I enjoy capture-the-flag competitions and occasionally document my walkthroughs. My writeups cover challenges from platforms like TryHackMe, and various CTFs.
- TryHackMe Room walkthroughs & notes
- Gajabaar Gajabaar CTF prep challenges
- Sudo Override Sudo Override challenge solutions
All writeups at writeups.aavashchhetri.com.np →
Education
Contact