Research Experience

My research focuses on reliable and context-aware machine learning for natural language processing, large language models, AI for cybersecurity, sensing systems, and biomedical AI. I study how models behave when data are complex, noisy, shifted, incomplete, or tied to real-world decision making. Detailed descriptions of selected research projects are provided on the Notable Projects page.

Research Themes
Reliable NLP, AI for Cybersecurity, Applied AI, and Mobile Sensing
Reliable NLP Distribution shift, calibration, selective prediction, uncertainty estimation, and human-in-the-loop decisions.
Context-Aware Sentiment Contrast, irony, discourse, aspect-level opinion, and domain shift.
AI for Cybersecurity Vulnerability severity prediction, CWE enrichment, security text analytics, and AI-assisted decision support.
Speech and Audio AI Speech representation diagnostics, task-relevant layer disruption, and deepfake audio analysis.
Sensing Systems Smartphone-based vibration sensing, mobile sensing, and structural condition monitoring.
Biomedical AI UWB sensing, non-invasive monitoring, and machine learning for health applications.

Research Experience by Academic Institution

Auburn University
Graduate Research Assistant | Department of Computer Science and Software Engineering | 2022–Present

My Ph.D. research at Auburn University develops reliable and context-aware AI methods for language understanding, cybersecurity decision support, speech representation analysis, intelligent transportation, and sensing systems. The main dissertation direction studies sentiment analysis when text contains linguistic complexity such as contrast, irony, implicit meaning, aspect-specific opinion, and distribution shift.

This research combines transformer-based modeling, uncertainty estimation, graph learning, multi-task learning, hierarchical evidence modeling, and task-aware evaluation. My goal is to build methods that remain useful beyond clean benchmark settings and can support practical decisions in real-world computing environments.

ShiftAware-Select Reliable NLP classification under distribution shift with calibration, uncertainty estimation, and abstain-to-human decisioning. Under review: IEEE ICMLA 2026.
CAGLSAv2 Position-invariant discourse-span attention for contrastive and length-robust sentiment analysis. Under review: IEEE ICMLA 2026.
Irony-Aware Twitter Sentiment Analysis Multi-task learning framework for sentiment, irony, and emotion understanding in social media text. Submitted: IEEE AIBThings 2026.
Hierarchical Severity Classification Cybersecurity vulnerability severity prediction with automatic CWE enrichment and sentence-level evidence. Submitted: IEEE AIBThings 2026.
DualGraph-RoBERTa Aspect-based sentiment analysis using syntactic dependency graphs and semantic co-occurrence evidence. Submitted: IEEE AIBThings 2026.
LayerLens Task-relevant layer disruption analysis for self-supervised speech encoders under noise, reverberation, and enhancement artifacts. Under review: IEEE/ACM TASLP.
Traffic Data Imputation Study Downstream evaluation of reconstruction accuracy and forecasting utility for intelligent transportation data. Under review: IEEE T-ITS.
Vi-Detect Smartphone-based vibration sensing for fine-grained component looseness detection. Poster published at ACM SenSys; full article under minor revision at ACM TOSN.
DeepFakeDetect Fine-grained deepfake audio detection using mechanical modeling of vocal folds. Under review: Journal of Information Security and Applications.
Discourse-Aware Domain Adaptation Planned sentiment transfer research using discourse structure and span-level evidence for domain-adaptive language understanding.
Reliable NLP Sentiment Analysis AI for Cybersecurity Speech Processing Sensing Systems Transformer Models Graph Learning Uncertainty Estimation
Universiti Malaysia Pahang
Graduate Research Assistant | M.Sc. by Research in Electrical and Electronic Engineering | 2018–2021

My graduate research at Universiti Malaysia Pahang focused on applied machine learning, digital signal processing, ultra-wideband sensing, and biomedical AI. I also collaborated on applied deep learning projects involving image-based disease detection, object detection, and breast cancer detection applications.

Non-Invasive Blood Glucose Estimation Machine learning with ultra-wideband signals for glucose-level estimation.
Breast Cancer Detection Support Collaborative AI-assisted analysis using ultra-wideband sensing concepts for early breast cancer detection.
Grape Leaf Disease Detection Deep learning for agricultural image-based disease recognition.
Palm Fruit Object Detection Deep transfer learning using YOLO-based object detection from plantation video data.
Land Deformation Analysis Deep learning support for deformation-related monitoring and applied sensing systems.
Machine Learning Deep Learning Digital Signal Processing UWB Technology Biomedical AI Applied Sensing
Jashore University of Science and Technology
B.Sc. in Computer Science and Engineering | Undergraduate Research Foundation

My undergraduate training at Jashore University of Science and Technology provided the foundation for my later research in machine learning, software systems, cybersecurity, and applied AI. I built core preparation in programming, algorithms, database systems, networking, operating systems, and software development.

My undergraduate thesis focused on ransomware file encryption attacks and data protection. This early work helped shape my interest in cybersecurity, software systems, and practical computing problems.

Ransomware File Encryption Attacks Undergraduate thesis on ransomware behavior, file encryption attacks, user-level security weaknesses, and data protection strategies.
Research Preparation Early preparation for graduate research through technical writing, implementation, problem solving, and applied computing projects.
Computer Science Cybersecurity Programming Algorithms Software Systems Research Foundation