During my time at the Security and Software Engineering (S2E) lab, I collaborated with Professor Joanna C. S. Santos at the University of Notre Dame on a research project titled Using Large Language Models to Generate JUnit Tests: An Empirical Study. Under her guidance, my role involved working on data preprocessing, generating outputs, and offering insights that contributed to the success of our project. Currently, our paper has been accepted for presentation at the 28^th International Conference on Evaluation and Assessment in Software Engineering (EASE 2024), esteemed as a core-A conference in the field. The paper quickly captured attention, amassing over 65 citations in a short time. In the project, I explored three generative models (Codex, GPT-3.5-Turbo, and StarCoder) to understand their capabilities in generating test cases and also looked at factors like compilation rates, test correctness, coverage, and test smells. This exploration and mentorship provided valuable insights into code-generation models which refined my research skills and broadened my perspective. I am eager to apply these insights and skills to future endeavors.

Under the guidance of Professor Dr. M. Sohel Rahman at BUET, I contributed to an undergraduate research project focused on steganography. My work, shaped by Dr. Rahman’s mentorship, centered on Enhancing the RGB Image Selection Process Using Genetic Algorithms for Steganography. This work allowed me to leverage my adaptability and understanding of complex concepts, particularly in genetic algorithms and deep learning. Our project aimed to improve the compatibility and imperceptibility of cover images in RGB image steganography through genetic algorithm-based methods. We conducted rigorous evaluations of various genetic algorithms, including tournament selection, elitism, and mutations, as well as assessment techniques like PSNR and SSIM. Additionally, we designed chromosomes for RGB images and incorporated the Modified Linear Congruential Generator (LCG) algorithm to achieve better results. The results demonstrated enhanced stego images, which are images with hidden data embedded within their pixel values, with reduced distortion, highlighting the method’s potential in secure communication.

Under the mentorship of Professor Dr. M. Sohel Rahman at BUET, I am deeply engaged in researching the virtual reality domain. Our project, Comprehensive Analysis and Detection of Interpersonal Rapport in Virtual Reality Environments, investigates the nuanced interactions within VR settings. We have extensively explored various facets of VR and AR, aiming to understand and enhance interpersonal rapport in these immersive environments. As part of this research, my team and I developed an application in Unity, which serves as a platform for conducting our experiments. We are currently conducting surveys to gather data and validate our findings. This research not only advances our knowledge of VR interactions but also fuels my fascination with the VR domain. I am eager to further explore this area and contribute to the development of more intuitive and effective VR systems in the future.

In collaboration with Professor Dr. Tanzima Hashem at BUET, I am working on a project focused on enhancing fairness in the MIMIC-IV Pipeline, a novel accessible electronic health record dataset. This research aims to address biases in healthcare data processing by developing a customizable pipeline for extracting, cleaning, and preprocessing MIMIC-IV data. Our goal is to improve the fairness metric known as Average Absolute Odds Difference (AAOD) by implementing various pre-processing, in-processing, and post-processing methods. This project involves meticulous data handling and sophisticated algorithmic adjustments to ensure equitable outcomes across diverse patient demographics. Through this research, I am contributing to the advancement of fair and unbiased healthcare data analysis, with the ultimate aim of improving patient care and health outcomes.

In collaboration with Dr. Md. Shohrab Hossain from BUET and Suryadipta Majumdar from CIISE, I worked on an innovative project focused on Binary Static Analysis in the Layer Concept of Docker. This project aimed to enhance the security of Docker containers by analyzing binaries within the layered architecture of Docker. Under the guidance of Professor Shoharab, I gained proficiency in using advanced binary analysis tools such as Ghidra and IDA Pro. I developed robust pipelines that integrate various automated vulnerability analysis tools, ensuring a comprehensive assessment of potential security threats. Additionally, I explored Docker's intricate layer concepts, furthering my understanding of containerization security. This collaborative effort highlighted my ability to work seamlessly in a team and my commitment to advancing research in cloud environments and reverse engineering. I am optimistic about the future outcomes.