Ph.D. in Information Systems
Jun 2026 – PresentUniversity of Maryland, Baltimore County (UMBC), Baltimore, MD, USA
About me
Seeking Summer 2027 research internships in computer vision, federated learning, LLM reasoning, and digital twins. My background fits Research Scientist, Applied Scientist, and ML Research Intern roles, and I am always open to research collaborations. Email me.
I am a Ph.D. student at the University of Maryland, Baltimore County, advised by Dr. Sanjay Purushotham. Currently, I am working on federated reasoning LLMs.
I have over three years of combined academic and industry experience. Previously, I worked as a Graduate Research Assistant at PIC Lab, Kumoh National Institute of Technology, South Korea, where I was advised by Dr. Taesoo Jun and contributed to research in industrial AI, multimodal learning, and warehouse automation. My research has been published in Journal of Information Security and Applications, IEEE Access, ICT Express, ICRA, ICC, ICMIC, and ICTC.
I also worked as a Software Engineer at Divine IT Ltd., where I developed enterprise software solutions and improved system performance through backend optimization and scalable software design.
News
-
Received an Honorable Mention at the UMBC IS Research Showcase for my poster A Study on Federated Fine-Tuning of Reasoning LLMs across Heterogeneous Clients.
-
Received the Best Paper Award at IEEE ICMIC 2026 for Adaptive Reconstruction and Deformable Attention for UAV Small-Object Detection.
-
FMACNN, a federated multi-attention CNN for artificial image detection, published in the Journal of Information Security and Applications.
-
Started my Ph.D. in Information Systems at UMBC as a Graduate Research Assistant.
-
AI-Driven Digital Twin Framework for Warehouse Logistics Automation Optimization appeared at an IEEE ICRA 2026 workshop in Vienna.
-
Received the Outstanding Graduate Research Achievement Award from Kumoh National Institute of Technology.
-
Two journal papers published: Stable Diffusion-augmented industrial anomaly detection in IEEE Access, and a review of AI-enhanced digital twins for warehouse logistics in ICT Express.
-
Began a Ph.D. in Electrical Engineering at Georgia Southern University, working on tiny-object detection for UAV imagery (later transferred to UMBC).
-
Completed my M.Eng. in IT Convergence Engineering at Kumoh National Institute of Technology with a 4.44/4.50 GPA.
-
Journal paper on blockchain-aided intrusion detection for marine tactical networks in The Journal of the Korean Institute of Communications and Information Sciences, plus two papers at the KICS Fall Conference 2025.
-
Two papers at IEEE ICTC 2025 on digital-twin-driven knowledge distillation and federated intrusion detection.
-
Released the RealAIGI dataset of real and AI-generated images at IEEE ICMIC 2025.
-
Best Paper Award at the KICS Summer Conference 2025.
-
BLIND-Twin, an LLM-based decision-support system for digital-twin-enhanced industrial AIoT, at IEEE ICC 2025.
-
TMNet, a transformer-fused framework for emotion recognition from EEG and speech, published in ICT Express.
-
Federated fine-tuning of Qwen-2.5-7B on finance data ranked 5th on the official Flower LLM leaderboard.
-
Two Best Paper Awards at the KICS Fall Conference 2024 (Gumi Industrial Complex Energy Self-Sufficiency Data Academic Paper Contest).
-
Three papers at IEEE ICTC 2024, including federated semi-supervised digital twins for human-machine interaction.
-
Joined Kumoh National Institute of Technology, South Korea, for my M.Eng. and research in the PIC Lab.
Selected Publications
Publications
Google Scholar131citations 7h-index 4i10-index Google Scholar, October 2026
Show all 40 publicationsHide publications
No publications match that filter.
Journal Articles
-
IEEE Access 2026
Stable Diffusion-Augmented Industrial Anomaly Detection Using Attention-Enhanced Multi-CNN Transformer Fusion
IEEE AccessSCIE · IF 4.2 · Q1
-
JISA 2026
FMACNN: Federated Multi-Attention CNN Framework for Artificial Image Detection
Journal of Information Security and ApplicationsSCIE · IF 4.4 · Q1
- ICT Express 2026
-
ICT Express 2025
TMNet: Transformer-Fused Multimodal Framework for Emotion Recognition via EEG and Speech
ICT ExpressSCIE · IF 4.6 · Q1
-
J-KICS 2025
Blockchain-Aided Intrusion Detection in Marine Tactical Network Using Reinforcement Learning
The Journal of the Korean Institute of Communications and Information Sciences
International Conference & Workshop Papers
-
ICMIC 2026
Adaptive Reconstruction and Deformable Attention for UAV Small-Object Detection
IEEE ICMIC, 2026
PaperCode coming soon
Best Paper Award
-
ICRA WS 2026
AI-Driven Digital Twin Framework for Warehouse Logistics Automation Optimization
IEEE ICRA Workshop, 2026
- ICAIIC 2026
- ICTC 2025
- ICMIC 2025
- ICC 2025
- ICMIC 2025
- ICCIT 2025
- ICTC 2025
- ICTC 2024
- ICTC 2024
- ICTC 2024
- ICCIT 2023
- ICCIT 2022
Under Review & In Preparation
-
Major Revision
Hierarchical Multi-Target Planning and Control for Autonomous Forklift Navigation with Quantum-Inspired Optimization and Reinforcement Learning
International Journal of Control, Automation, and Systems
-
Under Review
BELL-IDS: Blockchain-Empowered Large Language Model-Aided Context-Aware IDS in IoFT
IEEE Transactions on Network and Service Management
-
Under Review
Adaptive Quantum-Resistant Deep Reinforcement Learning Framework for 6G Healthcare IoMT Security
IEEE Journal of Biomedical and Health Informatics
-
Under Review
HybridQuantumNet: Hybrid Quantum–Classical Federated Learning for IoMT Intrusion Detection
International Journal of Machine Learning and Cybernetics
-
Under Review
Deep Q-Network Agent-Driven Dynamic Slotting Optimization in Digital Twin-Based Smart Warehouse Logistics
International Journal of Control, Automation, and Systems
-
Under Review
FAR-TOD: Failure-Prior Adaptive Reconstruction Guided Tiny Object Detection
WACV Workshop on Real-World Surveillance (RWS), 2027
- Submitted
Domestic conference papers (KICS) 14
-
KICS 2026
Privacy-Preserving Healthcare Data Governance Using Blockchain and Federated Learning
KICS Summer Conference, 2026
-
KICS 2025
Multi-modal Real-time Hazard Detection for Autonomous Driving Systems
KICS Fall Conference, 2025
-
KICS 2025
FlexSim-Based Digital Twin Framework for Smart Warehouse Monitoring and Performance Optimization
KICS Fall Conference, 2025
-
KICS 2025
Detecting Threats in Edge IoT Networks Using Federated Learning and Digital Twin
KICS Summer Conference, 2025
Best Paper Award
-
KICS 2025
Novel Zero-Day Zero-Touch Intrusion Detection System Using Generative-AI and Large Language Model
KICS Summer Conference, 2025
-
KICS 2025
XAI-Powered Digital Twin for Industrial Energy Prediction and Optimization
KICS Summer Conference, 2025
-
KICS 2024
DCFL-Chain: Digital-Twin-Based Collaborative FL-Integrated Energy Consumption Prediction for Smart Factory
KICS Fall Conference, 2024
Best Paper Award
-
KICS 2024
Energy-Twin: Explainable AI-Driven Energy Optimization using Digital-Twin for Smart Industry
KICS Fall Conference, 2024
Best Paper Award
-
KICS 2024
BFL-Net: Blockchain-Assisted Energy Prediction Based on FL-Integrated LLM for Industry 4.0
KICS Fall Conference, 2024
-
KICS 2024
Quantum Encrypted Collaborative Energy Consumption Demand Response Management for Smart Industrial System
KICS Fall Conference, 2024
- KICS 2024
-
KICS 2024
PureParking: A Decentralized, Secure Framework for Parking Space Sharing Using Blockchain
KICS Summer Conference, 2024
-
KICS 2024
A Blockchain-Based Framework for Distributing and Managing CNN-Derived Brain Tumor Detection Models
KICS Summer Conference 2024
-
KICS 2024
An Efficient, Traceable, and Transparent Maritime Logistic System Based on Blockchain
KICS Summer Conference, 2024
Honors & Awards
- 2026Honorable Mention, IS Research Showcase (November 2026), UMBC
- 2026Best Paper Award, IEEE ICMIC 2026
- 2026Outstanding Graduate Research Achievement Award (Spring 2026), Kumoh National Institute of Technology
- 2025Best Paper Award, KICS Summer Conference 2025 · Certificate
- 20255th place, Flower LLM leaderboard (finance), with federated fine-tuning of Qwen-2.5-7B
- 2024Two Best Paper Awards, KICS Fall 2024 Gumi Industrial Complex Energy Self-Sufficiency Data Academic Paper Contest · Certificates
- 2022Champion, Intra-University Programming Contest (IUPC) 2022, IUBAT · Certificate
- 2022Second Runner-up, UITS IUPC 2022 · Certificate
- 2022Ranked in BUET, RUET, BUBT, and AUST Inter-University Programming Contests · Certificates
- 2021Champion, Intra-University Programming Contest (IUPC) 2021, IUBAT · Certificate
- 2021Finalist, ACM International Collegiate Programming Contest (ICPC), Dhaka Regional · Certificate
- 2020First Runner-up, Intra-University Programming Contest, IUBAT
- 2019Best Fresher Programmer, Intra-University Programming Contest, IUBAT
- OngoingSolved more than 1,600 data-structure and algorithm problems on online judges · Profile
Education
Ph.D. in Electrical Engineering (Transferred)
Jan 2026 – May 2026Georgia Southern University (GSU), Statesboro, GA, USA
M.Eng. in IT Convergence Engineering
Mar 2024 – Dec 2025Kumoh National Institute of Technology (KIT), Gumi, South Korea
B.Sc. in Computer Science and Engineering
Jan 2019 – Dec 2022International University of Business Agriculture and Technology (IUBAT), Dhaka, Bangladesh
Experience
Graduate Research Assistant
Jun 2026 – PresentUniversity of Maryland, Baltimore County, Baltimore, MD, USA
- Building a framework to resolve reasoning collapse in federated LLMs under heterogeneous data distributions.
- Developing a mosquito-movement digital-twin simulation in NVIDIA Omniverse Isaac Sim, grounded in a literature survey of mosquito movement.
Graduate Research Assistant
Jan 2026 – May 2026Georgia Southern University, Statesboro, GA, USA
- Developed AR-TOD, an adaptive reconstruction-guided small-object detection framework for UAV imagery that reaches 33.1 APvt, exceeding full-grid reconstruction while using 83% fewer tiles and 61.6% fewer GFLOPs.
- Worked on depth estimation for autonomous-vehicle hazards and on BlueBoat surface-vessel setup and calibration.
Graduate Research Assistant, PIC Lab
Mar 2024 – Nov 2025Kumoh National Institute of Technology, Gumi, South Korea
Real-Time Autonomous Driving Hazard Analysis
Sep 2025 – Present- Developing a multimodal model that fuses speed, vision, and sound to detect driving hazards.
- Developed a lightweight regression model for speed- and sound-based driver-state detection, and a VLM for visual detection.
AI for Digital Twin-based Warehouse Logistics Optimization
Jul 2024 – Present- Developed a warehouse digital twin in NVIDIA Omniverse Isaac Sim and FlexSim to generate synthetic data.
- Developed a DQN agent that reduces average picking time by 30.83% and average travel distance by 13.15%.
IEEE ICRA Workshop 2026IEEE ICAIIC 2026ICT Express 2026KICS Fall 2025Code
Daycare Face and Action Recognition on Edge Device
Sep 2024 – Dec 2024- Developed a Transformer model for daycare face recognition with MTCNN facial embeddings (98% accuracy).
- Reached 97% accuracy in daycare action recognition with a BiLSTM model on MoveNet pose embeddings.
- Constructed a Korean children's face and action recognition dataset.
Industrial Manufacturing System Anomaly Detection
Mar 2024 – Jun 2024- Developed an attention-enhanced Multi-CNN Transformer fusion model with 98% accuracy.
- Used Stable Diffusion to generate synthetic defect images, easing data limitations and class imbalance.
- Improved baseline accuracy by 4.64% through optimized preprocessing, augmentation, and fine-tuning.
Data Scientist, AIMS Lab
Oct 2023 – Dec 2023United International University, Dhaka, Bangladesh
- Analyzed heart failure, kidney disease, and stroke data from rural areas of Bangladesh.
- Identified high-impact regions and uncovered reasons for disease prevalence.
Software Engineer (Python)
Sep 2022 – Sep 2023Divine IT Ltd., Dhaka, Bangladesh
- Developed APIs, modules, and user-facing features for the PrismERP, PrismHRM, and PrismATS software suites.
- Improved stability and performance by resolving bugs and optimizing with parallel processing, efficient algorithms, and database optimization.
- Contributed to full-stack development of Enterprise Resource Planning (ERP) and Applicant Tracking System (ATS) products.
Selected Projects
Distributed Financial LLM
Federated instruction-tuning of Qwen-2.5-7B on a finance corpus with Flower, LoRA via PEFT, and FedAvg across 50 IID client partitions, trained on 2×RTX 3090 GPUs.
FIQA/FPB/TFNS avg 61.53 · 5th on the Flower leaderboard
- Flower
- LoRA
- PEFT
Federated Artificial Image Detection
A federated multi-attention CNN that detects AI-generated imagery across IID and non-IID clients, with weighted FedAvg, percentile clipping, and divergence monitoring for stable convergence.
RealAIGI 10k images incl. Sora · 100 clients
- PyTorch
- FedAvg
- Attention
Multimodal Emotion Recognition
A transformer-fused recognizer that denoises and synchronizes speech and EEG streams (MFCC features, FastICA filtering, dynamic time warping), with CNN-BiLSTM and BiGRU encoders joined by cross-attention.
97% accuracy · ternary affect classification
- EEG
- Speech
- Cross-attention
RL-based IoMT Intrusion Detection
A lightweight, cost-sensitive intrusion detection system for IoMT using CNN–LSTM and contextual-bandit PPO, with top-K Random Forest feature selection to cut false negatives at the edge.
97.99% accuracy · 1.2 ms latency · <0.15 MB model
- CNN–LSTM
- PPO
- Edge
Gumi Industrial Energy Forecasting
Visualized real industrial-complex energy data, applied smoothing and scaling, and built a lightweight CNN-BiLSTM forecaster.
98% forecasting accuracy
- CNN-BiLSTM
- Time series
Chronic Kidney Disease Risk Prediction
A risk-scoring pipeline that cleans noisy CKD lab records and trains an XGBoost classifier, served through a FastAPI service with schema validation, a simple frontend, Docker, and Gunicorn.
>90% AUC on the held-out set
- XGBoost
- FastAPI
- Docker
Service & Mentoring
- Peer reviewer
- IEEE Access·ICT Express·Future Generation Computer Systems
- Mentoring
- Data Structures, Algorithms & Programming Language Mentor at IUBAT: beginner C and C++ sessions, intermediate data-structures and algorithms sessions, and competitive-programming learning sessions.
- Organizing
- Organizer, IUBAT Intra-University Programming Contest, 2021 and 2022
Technical Skills
- Data science
- TensorFlow, PyTorch, Scikit-Learn, Pandas, Keras, NumPy, Matplotlib, OpenCV, Power BI, LangChain, LangGraph
- Concepts
- Data analysis, multimodal AI, data visualization, foundation models, RAG, computer vision, federated learning, LLMs, generative AI, diffusion models, time-series forecasting, quantization, signal processing, intrusion detection, VLMs
- Software development
- Python, C++, OOP, SQL, data structures and algorithms, DRF, FastAPI, JavaScript, Express.js, RESTful APIs, MySQL, SQLite, NoSQL, Postman, Docker, Linux, AWS
- Simulation
- NVIDIA Omniverse Isaac Sim, FlexSim
Open to Research Collaboration
I am glad to hear from researchers and students working on related problems, especially:
- Federated learning
- LLM reasoning
- Small-object detection
- Vision-language models
- Multimodal AI
- Digital twins
- Reinforcement learning
- Generative AI
- Edge AI
- Intelligent automation
Contact
- malam4@umbc.edu
- Department
- Information Systems
- Institution
- University of Maryland, Baltimore County
1000 Hilltop Circle, Baltimore, MD 21250, USA - Profiles
- Google Scholar · GitHub · LinkedIn · ResearchGate · Kaggle
