About Me
I am a Ph.D. candidate in Computer Science and Engineering at The University of Texas at Arlington, advised by Dr. Yuede Ji. My research focuses on the security and trustworthiness of AI systems, particularly AI supply-chain security, model provenance, and vulnerability propagation across base models and their derivatives.
I use graph-based methods, static analysis, and large-scale empirical evaluations to understand how AI models are built, adapted, and connected—and how security risks travel through those relationships. I also study prompt injection and cross-agent spillover in autonomous and multi-agent AI systems, alongside trustworthy, resource-efficient edge AI. My goal is to develop scalable, auditable, and secure AI systems throughout their development and deployment.
Before my doctoral studies, I worked in cybersecurity across banking, consulting, and national infrastructure in Bangladesh. My roles included Chief Information Security Officer at BRAC Bank PLC, National IT Security Consultant with e-GOV CIRT, and positions at Eastern Bank PLC, KPMG, and IBM. These experiences inform my approach to research: connecting practical security problems with rigorous analysis and usable defenses.
Recent Updates
2026
DiscoAct, our work on securing autonomous multimodal LLM agents against cross-modal trust flattening, is accepted at IEEE ICTAI 2026 (to appear).
AgentFuzz: Trace-Level Fuzzing for Prompt Spillover in LLM Multi-Agent Systems is a refereed conference paper at IEEE ICMLA 2026.
LiteAugNet, a lightweight semantic-guided augmentation network for efficient edge-level image classification, appears in ICPR 2026 (Springer LNCS).
Co-authored Beyond CVSS: Context-Aware Exploitability Assessment of Vulnerabilities in HPC Containers, published at IEEE Cloud Summit 2026.
SEGA-Net, our work on LLM-guided augmentation for low-resolution image classification, appears in IEEE AIxDKE 2026.
2025
HuggingGraph: Understanding the Supply Chain of LLM Ecosystem appeared at ACM CIKM 2025, with open research artifacts supporting reproducibility.
Completed a Summer Security Audit Internship at the Federal Home Loan Bank of Dallas.
Named Teaching Assistant of the Month in UTA's Department of Computer Science and Engineering.
Ongoing
- 2024–present
Serving as a peer reviewer for IEEE OJ-COMS, ICMLA, ICTAI, ICPR, and AIxDKE.
Current work includes ProvCode, ChainInfer, and SEGA-COM (under review), alongside ongoing research on AI supply-chain vulnerability propagation and model lineage. See my CV for publication details and research status.