Madan Baduwal

PhD in CSE | Mississippi State University

Jan 2025 - Present

PhD in Computer Science & Engineering at Mississippi State University. Graduate Research Assistant working on Machine Learning and Wireless Communications, with broader interests in Physics, English, Mathematics, Computer Science, and Robotics.

Jan 2023 - Dec 2024

Graduate Research Assistant and Student Assistant Editor at University of Texas Permian Basin under Prof. Quan Yuan. Researched polyp segmentation and multimodal AI by fine-tuning models including YOLOv8, U-Net, Detectron, CLIP, ViLBERT, LXMERT, BLIP, and LLaVA, curated university-ready text-image datasets, and supported publication editing for the UTPB Journal of Undergraduate Research and UG Research Book.

Jan 2022 - Dec 2022

Senior Machine Learning Engineer at Fusemachines. Led development of Matrice.ai, a no-code data-centric AI platform that cut deployment time by 40% and development cost by 80%, built ML SaaS products for 5+ hospitals, and automated otoscopy image segmentation that generated $100k annually.

Jan 2021 - Dec 2021

Computer Vision Engineer (R&D) at National Innovation Center. Built vision and navigation systems for service robots, improving edge inference and pathfinding, including collaboration with Mahabir Pun.

Feb 2020 - Feb 2022

Machine Learning Engineer at Fusemachines. Led AI systems for education and phishing detection, supported large-scale deployments, and delivered workshops, trainings, and machine learning content for thousands of learners.

Aug 2019 - Feb 2020

Software Engineer Intern at Omnibluetech. Built Django APIs, AWS background workers, and web apps for document processing, retail, and consultancy systems.

2015 - 2019

BE in Computer Engineering at Tribhuvan University, Institute of Engineering, working under Prof. Sanjeev Prasad Panday. This is also where I built my foundation in AI, software engineering, mathematics, and vision robotics.

Selected Publications

Lightweight Multimodal Radar Interference Detection at Low SINRs in CBRS

Madan Baduwal, Vini Chaudhary

IEEE International Conference on Communications (ICC), 2026

A multimodal machine learning framework that fuses in-phase/quadrature (IQ) samples and spectrogram representations for radar interference detection in CBRS under challenging low-SINR conditions. The framework combines quantization-aware training and SHAP-based feature pruning to reduce computational complexity while maintaining at least 99% detection accuracy, achieving low-latency inference for lightweight Environmental Sensing Capability (ESC) sensors.

Publication Bibtex

PERFECT: Personalized Federated Learning for CBRS Radar Detection

Shafi Ullah Khan, Madan Baduwal, Vini Chaudhary, Debashri Roy

35th International Conference on Computer Communications and Networks (ICCCN), 2026

A personalized federated learning framework for CBRS radar detection across geographically distributed Environmental Sensing Capability (ESC) sensors. PERFECT preserves privacy through local model training and addresses non-IID data through ESC-level personalization, achieving the mandated 99% radar detection recall while improving privacy, efficiency, and scalability for dynamic spectrum sharing.

Paper Bibtex

Who Ran My Circuit: Calibration Data-Based Fingerprinting of Quantum Cloud Hardware

Wilson Patterson, Madan Baduwal, Vini Chaudhary

35th International Conference on Computer Communications and Networks (ICCCN), 2026

A calibration data-based quantum device fingerprinting framework for verifying which quantum processor executed a workload in multi-tenant quantum cloud environments. The approach uses historical calibration data and XGBoost to identify IBM quantum processors, achieving 99.03% accuracy in binary identification and 82.85% accuracy in multi-class identification without requiring additional quantum circuit execution.

Publication Bibtex

Federated Learning: A Survey of Core Challenges, Current Methods, and Opportunities

Madan Baduwal, Priyanka Paudel, Vini Chaudhary

Computers, 2026

A survey of the main systems and learning challenges in federated learning, covering practical constraints, current methods, and open opportunities for future work.

Paper Bibtex

Signature Verification using Convolutional Neural Network and Autoencoder

Prakash Ratna Prajapati, Samiksha Poudel, Madan Baduwal, Subritt Burlakoti, Sanjeeb Prasad Panday

Journal of the Institute of Engineering, 2021

Early published work on writer-dependent signature verification using learned feature extraction with convolutional neural networks and autoencoders.

Paper Bibtex Code

Hybrid (Transformer+ CNN)-based Polyp Segmentation

Madan Baduwal

arXiv preprint, 2025

A medical imaging segmentation pipeline that combines transformer-style context modeling with convolutional detail recovery for more accurate polyp segmentation.

Paper Bibtex