Cauli-Det: Modified YOLOv8 for Cauliflower Disease Detection
ResearchComputer Vision
91.1% mAP across three cauliflower disease classes from smartphone-captured field images. Published in Frontiers in Plant Science.
ResearchComputer Vision
91.1% mAP across three cauliflower disease classes from smartphone-captured field images. Published in Frontiers in Plant Science.
ResearchAI Safety
Developing methods to formally verify safety properties of activation-editing techniques in transformer models, on toy models, aimed at a top-tier conference submission.
ResearchReinforcement Learning
Multiple RL agents independently observe local patches of a mammogram ROI, communicate, and reach a decentralized consensus classification. Presented at IEEE ICDABI 2023.
SoftwareAI Agent
Converts natural-language requests into SQL, executes them, and returns results conversationally — prompt engineering, RAG, vector databases, Gemini and OpenAI APIs.
Published in 2023 4th International Conference on Data Analytics for Business and Industry (ICDABI), IEEE, 2023
A cooperative multi-agent reinforcement learning framework where agents independently observe regions of mammogram images, communicate, and reach consensus for breast cancer classification.
Q1Scopus Indexed
Published in Frontiers in Plant Science, Vol. 15, 2024
A modified YOLOv8 architecture for multi-class cauliflower disease detection and localization from smartphone-captured images, achieving 91.1% mAP across three disease categories.
Scopus Indexed
Published in Data-Driven Clinical Decision-Making Using Deep Learning in Imaging (Studies in Big Data, Vol. 152), Springer, 2024
A generator–discriminator GAN architecture to synthesize breast-cancer ROI images, augmenting sparse mammogram datasets to support downstream classification model training.
Published:
Presented the multi-agent reinforcement learning framework for mammogram ROI classification described in the accompanying paper.
Undergraduate courses, American International University-Bangladesh (AIUB), 2024
Teaching programming, computer architecture, operating systems, software engineering, and web development to undergraduate Computer Science students. Dec 2024 – Present.
Lead Developer, Web Development & Automation Team — designing and building internal systems to digitize and automate departmental activities and ensure smooth, efficient operation.
Contributor, Course Registration & Admission Management Systems — contributed to the student course registration system supporting AIUB’s open-credit registration model, and to the admission management system.