Experience
Experience and skills →Publications
All publications in detail →-
Survey of 274 papers on LLM-agent applications, threats and defenses.
Supervisors: Dr. Md Rizwan Parvez, Dr. Farig Sadeque · Collaborators: Md Nafiu Rahman, Asif Shahriar
-
Real-time Chrome extension that flags secrets in GitHub issue reports.
Supervisors: Dr. Rifat Shahriyar, Dr. Gias Uddin · Collaborators: Zahin Wahab, Md Nafiu Rahman
-
LLM pipeline and 54,148-instance benchmark for secrets in GitHub issues.
Supervisors: Dr. Rifat Shahriyar, Dr. Gias Uddin · Collaborators: Zahin Wahab, Md Nafiu Rahman
-
Two collaborating LLMs generate code from Bangla descriptions with self-refinement.
Collaborators: Md Nafiu Rahman, Mahir Labib Dihan
-
Fine-tuned LLMs cut false positives when detecting secrets in code.
Supervisors: Dr. Rifat Shahriyar, Dr. S M Sohan · Collaborators: Zahin Wahab, Md Nafiu Rahman
-
Under review: ARRIgnored, Not Refused: What Low Prompt-Injection Success Rates Hide in Small Tool-Using Agents
Low prompt-injection success rates in small agents mostly reflect ignored attacks, not refusals.
Supervisors: Dr. Farig Sadeque · Collaborators: Ariyan Hossain · Students supervised: Shabnam Sharmin Tarin, Synthia Akter Promi, Kazi Shahid Ahmed Galib, Farhan Zarif
-
Under review: ARRSeeing Through the Forgery Is Not Enough: Regressive Sycophancy under Fabricated Evidence in Medical Vision-Language Models
Medical VLMs abandon correct answers when users present fabricated prescriptions.
Supervisors: Dr. Chowdhury Mohammad Rakin Haider, Dr. Rifat Shahriyar · Collaborators: Soumya Swagata Biswas
-
Under review: ICLR 2027AgentTell: Behavioural Side-Channel Leakage in Browser-Use Agents
Browser agents leak private facts through their actions across websites.
Supervisors: Dr. Md Rizwan Parvez, Dr. Farig Sadeque · Collaborators: Md Nafiu Rahman, Asif Shahriar
About Me
I am a Lecturer in the CSE Department at BRAC University and a Research Assistant at BUET, where I completed my B.Sc. in 2025. My research applies large language models to software engineering and security problems and tests where these models fail. My work has been accepted at ESEM, MSR, FSE, TMLR and the BLP workshop at IJCNLP-AACL. After I was selected for the UIUC++ Summer Internship in 2026, I did a remote internship under the supervision of Dr. Wenxi Wang at the University of Virginia, with whom I still collaborate on software engineering research. I teach three courses: Computer Architecture, Software Engineering and System Analysis and Design.