Monirul I Mahmud

Chapter So Far

My philosophy with research is the same way a non-Newtonian fluid reacts to pressure: fluid and adaptable when exploring new ideas, but rock-solid when tackling complex problems. As a researcher, transitioning from my MSc at Fordham University to my Ph.D. at the University of Cincinnati, my core focus lies at the intersection of Multimodal AI, Efficient LLM, VLM, and differential privacy. My goal isn't just to build highly accurate models, but to design sustainable, efficient and privacy-aware systems that cross disciplinary boundaries. On the technical side, my skills include Python, TensorFlow, and PySpark, as well as, D3.js and Tableau to bring data to life. Outside the academic and professional pursuits, you can usually find me playing football, hiking, traveling, or geeking out over the latest space innovations. Always open to connecting with fellow researchers and interdisciplinary thinkers!

Latest Updates

SEP 2026

Fresh out of the oven! Our new Q1 paper just dropped in the Annals of Data Science. doi.org/10.1007/s40745-026-00718-z

AUG 2026

Excited to share my new IEEE Transactions paper in IEEE TCE (IF: 10.9), introducing "ReMerkle: Proof-of-Reputation-Driven Compact Merkle Verification for IoMT." doi.org/10.1109/TCE.2026.3723244

APR 2026

Proud to announce my first IEEE Transactions paper in IEEE TCE, on "Reputation-driven Federated Learning for secure & consumer-centric IoMT." doi.org/10.1109/TCE.2026.3685994

Feb 2026

I'm excited to share that I’ve been accepted to the PhD in Computer Science at the University of Cincinnati, starting Fall 2026!

Dec 2025

Another Q1 Journal Acceptance: Attention and Residual Mechanism-based CNN Architecture (ARC-Net) with Enhanced Fairness Generalization for Deepfake Facial Image Detection

SEP 2025

Our paper "DFU_DIALNet" is now online! DOI: https://doi.org/10.1371/journal.pone.0330669

AUG 2025

Our paper "DFU_DIALNet: Towards Reliable and Trustworthy Diabetic Foot Ulcer Detection with Synergistic Confluence of Grad-CAM and LIME" is accepted at PLoS ONE (Q1 Tier)!

DEC 2024

A proud milestone in my journey—I've been accepted into the M.S. in Analytics program at Georgia Institute of Technology! Very excited to join one of the world's Top 10 CS Ranked universities.

NOV 2024

Our paper "Linear Discriminant Analysis in Credit Scoring: A Transparent Hybrid Model Approach" has been accepted to IEEE International Conference on Computer and Information Technology (ICCIT) 2024.

Oct 2024

Our paper, titled "Optimizing Stroke Detection: An Analysis of Various Feature Selection Methods" has been published online (doi: 10.1145/3675094.3677602) at the UbiComp 2024 Conference, Australia.

MAY 2024

I am honored to serve as the Reviewer in peer-reviewed IEEE Pervasive Computing, Array, IEEE Internet of Things Journal.

Mar 2024

I have started my Masters in Computer & Information Science at Fordham University, New York with GSAS Centennial Scholarship.

DEC 2023

I am officially graduated from North South University with a Bachelors in Computer Science & Engineering.

Jul 2023

I have started working as Research Assistant at Design Inclusion & Access Lab (DIAL). Working on the project - 'Feasibility of Alternative Credit Scoring for the Credit Invincibles of Bangladesh' funded by NSU-CTRG (CTRG-23-SEPS-02)

Monirul Islam Mahmud AI security and privacy researcher

Interested in Large Language Model with Multimodal AI and Vision-Language Model

Machine Learning, NLP, LLM, VLM 95%

Blockchain, Security, Smart Contract90%

Python, PySpark, Pandas & Seaborn 80%

Power BI, Tableau & D3.js 70%

A/B Testing, Hypothesis Testing 65%

Education

University of Cincinnati, Ohio

Aug '26 - Present

Ph.D. in Computer Science & Engineering

➛ Serving as a Teaching Assistant (TA) for the Senior Design I course.
➛ Conducting research on "Calibrated Selective Attribution under Adaptive Threats" under the supervision of Professor Justin Zhan.

Fordham University, New York

2024 - 2026

M.S. in Computer & Information Science

➛ As a Full-time Graduate Student, 90% of my Tution Fee is funded by GSAS Centennial Scholarship.
➛ Developing expertise in areas including but not limited to Smart Contracts, Distributed Ledger systems, Trustworthy AI, Text Mining and Graph Mining. [Transcript] [Certificate]

North South University (NSU), Dhaka

2020 - 2023

B.Sc in Computer Science & Engineering

➛ Engaged in a comprehensive curriculum covering key areas such as Algorithms, Data Structures, Software Engineering, Theory of Computation and Machine Learning. [Transcript][Certificate]

Experience

Clinical Linguistics & Health Outcomes (CHO) Lab

2026 - Present

Research Assistant

➛ Working under Dr. Sunghye Cho on a 'STELLAR: Steering-Vector Enhanced LLM Agents for Realistic Digital Twins in Mental Health' and 'Personalized longitudinal markers of psychosis severity using automated speech and language analysis'. Find me

Fordham University

2025 - 2026

Research Assistant

➛ Working under Dr. Mohamed Rahouti on a 'Cryptographically Verifiable and Agent-Based Reputation Adjustment for Peer-to-Peer Platforms Rating' and 'Multimodal Jailbreaking and Adversarial Attacks on Vision-Language Models: A Survey' (Under Review, ACM CSUR).

Design Inclusion and Access Lab (DIAL)

2024 - 2025

Graduate Research Assistant

➛ Team Lead of Medical Imaging group - mentoring 5 members to ensure successful project execution under Dr. Nova Ahmed.
➛ Worked on 2 projects - 'Attention and Residual Mechanism-based Deepfake Recognition' and ‘DFU_DIALNet: Reliable and Trustworthy Diabetic Foot Ulcer Detection’ projects. Find me

Recent Projects

FairRate ride-sharing reputation project by Monirul Islam Mahmud

FairRate for Uber/Lyft

Fair Star Rating for Ride-sharing Platform's Driver

Vision-language model jailbreaking research by Monirul Islam Mahmud

VLM Jailbreaking

Survey on Adversarial Attacks on Vision-Language Models

Secure WBAN and IoMT federated learning project

Secure WBAN/IoMT System

Blockchain based FL with Proof of Reputation (PoR)

Trustworthy Keylogger detection

Ensemble and Feature Selections through Explainable AI

Hyperbolic CLIP cloud perception project

Hyperbolic CLIP for Cloud Perception

Open-Vocabulary Segmentation in Ground-Based Imagery

Deepfake image detection research project

Attention & Residual based DeepFake

ARC-Net with XAI for Deepfake Image Detection

Ground based Cloud Segmentation

U-Net based FCNN for sharper Sky insights

Merkle tree based credit scoring project

Markle Tree based Credit Scoring

Trustworthy Zero-Knowledge Proof Enhanced Approach

Human activity recognition wearable sensor project

Elderly People Activity Recognition

Multi-Head Temporal Attention based Bi-LSTM Approach

Diabetic foot ulcer detection DFU_DIALNet project

GradCam based Diabetic Foot Ulcer

Trustworthy proposed DFU_DIALNet to detect DFU

Stroke detection machine learning project

PCA & LDA based Stroke Detection

Linear Discriminant Analysis with XAI

Alternative Credit Scoring

Multi-Domain Validation of Bayesian Optimized Ensembles

Health Express BD

Emergency Ambulance, Medicine & Doctor.

NSU Study Portal project by Monirul Islam Mahmud

Student Study Portal

Definition, Books, Homework, Task.

LinkedIn Job Analysis 2024

Uncover trends in top industries, skills in USA

Benchmark Datasets

1. Human Activity Recognition from Wearable Sensors

A comprehensive dataset of 72,000 readings of 6 persons, collected using a 3D accelerometer and gyroscope sensor from Rasperry PI 4 Device. We also proposed a mobile app to collect and tracking human activity.

Dataset Link: (Coming soon)

2. Diabetic Foot Ulcer Dataset

A specialized image dataset comprising 500 foot images (250 normal, 250 with ulcers) for the detection of Diabetic Foot Ulcers. This dataset was collected with approval from the North South University IRB/ERC (Application No. 2024/OR-NSU/IRB/1110).

Dataset Link: Click Here

3. DeepFake and Real Face Image Collection

A dataset of 500 facial images, including real Bangladeshi individuals of diverse genders and ages, collected with explicit written consent for deepfake detection research. Data collection was approved by North South University IRB/ERC (Application No. 2024/OR-NSU/IRB/1109).

Dataset Link: Click Here

4. Ground-Based Cloud Segmentation Dataset

A segmented dataset of 110 ground-based cloud images, meticulously annotated using Roboflow. This dataset was developed for training and evaluating U-Net based Fully Convolutional Neural Networks (F-CNN) for cloud segmentation tasks.

Dataset Link: (Coming soon)

5. FairRate: Ride-Sharing Platforms Dataset

This benchmark dataset consists of 9,158 trip records from NYC TLC trip dataset (including real fares, distances, and weather) with synthetic data of 800 simulated riders. It features programmatically injected contextual biases and highly realistic, LLM-generated review text (Gemini 2.0).

Dataset Link: (Coming soon)

Publications

M. I. Mahmud, M. S. Reza, F. Elias, K. A. Ahmed, M. Ahammad, I. A. Abeer, and N. Ahmed. "Multi-Domain Validation of Bayesian Optimized Stacking Ensembles for Financial Credit Risk Modeling with Granular Explainability," Annals of Data Science. 2026. DOI: 10.1007/s40745-026-00718-z.
J. Bieniek, M. I. Mahmud, M. Rahouti, A. Chehri, and G. Jeon, "ReMerkle: Proof of Reputation-driven Federated Learning with Compact Merkle Verification for Consumer-centric IoMT," IEEE Transactions on Consumer Electronics. DOI: 10.1109/TCE.2026.3723244

M. I. Mahmud, J. Bieniek, M. Rahouti, A. Chehri, and G. Jeon. "Trustworthy Consumer-Centric IoMT: Reputation-Driven Federated Learning for Secure Medical Data Sharing," IEEE Transactions on Consumer Electronics. 2026. DOI: 10.1109/TCE.2026.3685994.

M. S. Reza, F. Elias, M. I. Mahmud, and N. Ahmed, "Attention and Residual Mechanism-based CNN Architecture (ARC-Net) with Enhanced Fairness Generalization for Deepfake Facial Image Detection," PLoS ONE. 2026. DOI: 10.1371/journal.pone.0340099.

M. I. Mahmud, M. S. Reza, M. O. A. Akash, F. Elias, and N. Ahmed. 2025. “DFU_DIALNet: Towards reliable and trustworthy diabetic foot ulcer detection with synergistic confluence of Grad-CAM and LIME,” PLoS ONE, vol. 20, no. 9, p. e0330669, USA. DOI: 10.1371/journal.pone.0330669..

M. I. Mahmud, M. S. Reza, and S. S. Khan. 2024. “Optimizing Stroke Detection: An Analysis of Different Feature Selection Approaches.” In Companion of the 2024 on ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp '24), Melbourne VIC, Australia, 142–146. DOI: 10.1145/3675094.3677602.

M. S. Reza, M. I. Mahmud, I. A. Abeer and N. Ahmed, "Linear Discriminant Analysis in Credit Scoring: A Transparent Hybrid Model Approach," 2024 27th International Conference on Computer and Information Technology (ICCIT), Cox's Bazar, Bangladesh, 2024, pp. 56-61. DOI: 10.1109/ICCIT64611.2024.11022149.

M. I. Mahmud, M. S. Reza, and H. Akter, “Human Activity Recognition Using Multiple Learning & XAI Techniques with Wearable Sensor Data,” North South University, Dhaka, Bangladesh. 2023. repository.northsouth.edu/handle/123456789/1093

M. I. Mahmud, “Towards Trustworthy Keylogger detection: A Comprehensive Analysis of Ensemble Techniques and Feature Selections through Explainable AI,” arXiv.org, May 22, 2025. https://arxiv.org/abs/2505.16103

M. I. Mahmud, “ZeroML: A Next Generation AutoML Language,” arXiv.org, May 23, 2025. https://arxiv.org/abs/2505.18243

Under Review ⏳

M. I. Mahmud*, M. S. Reza*, M. Rahouti, and J. Qadir. “Multimodal Jailbreaking and Adversarial Attacks on Vision-Language Models: A Survey.” ACM CSUR.
Jan Bieniek, M. I. Mahmud, Henok Wondimu, Dinesh C. Verma, Tom Sheffler and Mohamed Rahouti. 2026. “FairRate: A Practical and Fair Reputation Update Pipeline for Ride-Hailing Platforms.” IEEE Transactions on Intelligent Transportation Systems.
M. I. Mahmud, S. Patel, M. Rahouti, Kaiqi Xiong and A. Chehri, "Secure Credit Scoring: A Merkle Tree and Zero-Knowledge Proof-Enhanced Approach with Proactive Security Strategies in Consumer Applications," IEEE Transactions on Information Forensic & Security.
M. I. Mahmud, M. S. Reza and I. T. Naheen, "Multi-Head Temporal Attention Enhanced Bi-LSTM for Human Activity Recognition: Introducing a Novel Benchmark Wearable Sensor Dataset," Knowledge-Based Systems.