I am an undergraduate researcher in the Computer Science & Engineering department at the University of Dhaka. I am advised by Md. Shamsul Islam. I am seeking a PhD degree where I would become a leader in the NLP research community, working on the development of robust and transparent NLP/LLM systems for the diverse global community. I intend to carry out research in multilingual transfer learning, stance and sentiment analysis, as well as applied deep learning for social good, particularly for low-resource languages like Bangla.
Education
CGPA: 3.46/4.00 (through 7th semester)
Relevant coursework: Natural Language Processing, Machine Learning, Deep Learning, Data Structures & Algorithms, Probability & Statistics, Artificial Intelligence
Thesis: TerraSeg: Deep Learning-Based Multi-Class Segmentation of Satellite Imagery for Urban Spatial Analysis, supervised by Md. Shamsul Islam
Research
My research focuses on Machine Learning, Deep Learning, and Computer Vision, with an emphasis on developing robust and practical AI systems for real-world applications. Current and recent research threads include:
- Deep learning-based Bangladeshi mango variety classification using EfficientNetB0, achieving 98.01% accuracy and deployed as a real-time web-based prediction system (JBAS'26)
- Building segmentation and semantic land-cover mapping from high-resolution satellite imagery using advanced deep learning-based semantic segmentation techniques (undergraduate thesis)
- Road scene object detection in dense urban environments using an improved YOLOv8-based architecture, specifically optimized for real-world Dhaka city traffic scenarios
- Transfer learning and deep learning for computer vision, focusing on image classification, object detection, data augmentation, model evaluation, and deployment of practical AI applications
Selected Publications
Selected Projects
A deep learning-based mango classification system for identifying Bangladeshi mango varieties from images, achieving 98.01% accuracy and enabling real-time prediction through a web-based interface.
Developing a deep learning-based semantic segmentation system for building extraction and land-cover mapping from high-resolution satellite imagery, covering buildings, roads, vegetation, water bodies, paddy fields, and open land.
A robust multi-class road scene object detection system using an improved YOLO-based architecture, designed for dense and high-occlusion urban traffic scenarios using real-world Dhaka city road scenes.
A subscription-based eBook platform with Stripe payment integration that allows users to subscribe, access, and read books online through a web-based interface.
A GSM-based wireless notice board that enables real-time message display through SMS, allowing remote updates without requiring manual interaction with the display system.
An Arduino-based smart dustbin that automatically opens its lid when a user is detected nearby using an ultrasonic sensor and servo motor.
News
- 2026: Paper accepted at ECCT 2026
- 2026: Paper published in the Journal of Bangladesh Academy of Science, Vol. 50
- 2026: Paper on ideological stance detection submitted to IEEE SPICSCON 2026
- 2026: Selected as a Research Assistant, Department of Computer Science & Engineering, Faridpur Engineering College
- 2025: Started undergraduate thesis on Deep Learning-Based Building Segmentation and Semantic Land-Cover Mapping from High-Resolution Satellite Imagery
- 2025: Became Lead, FEC Road to Higher Education
Awards
Leadership, Mentorship & Community Engagement
Contributing to research activities in Machine Learning, Deep Learning, and Computer Vision, with a focus on semantic segmentation, image classification, object detection, and practical AI applications.
Leading initiatives focused on higher education opportunities, study abroad guidance, scholarship information, application preparation, and academic development for students of Faridpur Engineering College.
Engaged in undergraduate research on Deep Learning-Based Building Segmentation and Semantic Land-Cover Mapping from High-Resolution Satellite Imagery, while exploring advanced deep learning and computer vision techniques for real-world applications.
Solved 400+ programming problems across CodeChef, Codeforces, LeetCode, and Beecrowd, strengthening algorithmic problem-solving and computational thinking skills.
Technical Skills
PyTorch, TensorFlow, Keras, CNN, Transfer Learning, Data Augmentation, Model Evaluation
Image Classification, Object Detection, Semantic Segmentation, YOLOv8, OpenCV, EfficientNetB0
Python, C++, Java, C
Git, VS Code, Google Colab, Kaggle, Streamlit, Arduino IDE, Proteus
Django, Python, SQLite, MySQL, Stripe
Miscellaneous
Outside of research, I enjoy mentoring juniors in competitive programming and organizing coding bootcamps and hackathons.