Hafijul Hoque Chowdhury
I am Hafijul Hoque Chowdhury. I graduated from Bangladesh University of Engineering and Technology (BUET) with a B.Sc. in Computer Science & Engineering. I currently work as a Lecturer at United International University (UIU). My undergraduate thesis was on identifying and fixing user-level security issues in smart home devices using smaller LLMs, supervised by Dr. Shohrab Hossain and Dr. Suryadipta Majumdar.
At UIU, I mainly teach multiple theory & lab courses. Besides it, I supervise/co-supervise multiple groups of students on their Final Year Design Projects and term papers. Before this, I worked as a Software Engineer at Samsung Research & Development Bangladesh (SRBD), on the Samsung Health Research Platform (HRP) projects.
My undergraduate thesis introduced me to the field of security research, where I mainly used LLMs to address existing security and privacy problems.Building on this experience, my research interests lie in two complementary directions: using LLMs and machine learning to address security and privacy problems in software, systems, networks, and studying vulnerabilities in LLM and ML systems and developing methods to mitigate them.
News
-
August, 2025
Undergraduate Thesis titled "Identifying and Addressing User-level Security Concerns in Smart Homes Using “Smaller” LLMs" published at the 2025 Privacy, Security, and Trust (PST 2025) conference (Acceptance rate: 26%).
-
February, 2025
First runners-up in the IEEE Signal Processing Cup 2025. Team Straw Hats defeated over 100 teams from around the world at the DFWild Cup in deepfake detection.
Research Experience & Publications
Completed Work
Final Year Design Project Supervision
An ongoing final year design project I'm supervising. Using a synthetic paired-image dataset of Bangladeshi faces — same face, only the skin tone changed — we're testing whether LLMs hold colorism bias, a widespread but under-studied issue in South Asia. Instead of borrowing Western race categories that don't quite map onto this region, we're measuring it in a way that fits the South Asian context.
An ongoing final year design project I'm supervising. Most LLM tutors just answer whatever you ask, no matter how much you already know. This one doesn't. It tracks the learner's state, flags when someone's leaning on the AI a bit too much, and only then decides — through a separate pedagogical engine — how to teach, before the LLM ever writes a word. Built around Data Structures and Algorithms, where jumping straight to the answer usually means learning nothing.
Work Experience
My primary responsibility is teaching several courses each trimester. I am currently supervising two Final Year Design Projects (FYDPs) and co-supervising another. While teaching Machine Learning and Data Mining, I have also had the opportunity to supervise several promising term papers, many of which are now being prepared for submission.
- Spring 2026: Data Mining (Theory), Data Structures & Algorithms I (Lab)
- Fall 2025: Data Structures & Algorithms I (Theory & Lab), Machine Learning (Theory)
- Summer 2025: System Analysis & Design (Theory), Discrete Mathematics (Theory), Digital Logic Design (Lab)
Worked in the Samsung Health Research Platform (HRP) team, contributing to both Android (Kotlin) and backend (Spring Boot) development focusing on scalable service design and system efficiency.
- Android: Implemented real-time wearable status syncing, fixed critical issues in wearable & mobile apps, using Jetpack Compose and gRPC.
- Backend: Implemented key platform features with Spring Boot (Kotlin), MongoDB, and integrated cloud storage using Amazon S3.
Education
HONORS & AWARDS
1st Runners Up Team — IEEE Signal Processing Cup 2025
Presented at ICASSP 2025
Dean's List Award
BUET — Awarded in 3 out of 4 levels