Deep Learning Enthusiast
CSE graduate passionate about AI, machine learning, and computer vision, with hands-on experience in building datasets and developing AI projects.
I'm a Computer Science and Engineering graduate from Southeast University. I started out building my foundation in C, C++, and Java, which helped me understand how software works from the ground up. These days, I mostly work with Python, Machine Learning, and Deep Learning, with a growing interest in Computer Vision and AI research. I've worked on projects involving image datasets, Bangla Sign Language recognition, AI-generated image detection, and deepfake detection. I also enjoy creating and working with datasets, especially when they can be useful for AI research and real-world applications. I'm comfortable working across different areas of development, but Iām most interested in the logic, data, and intelligence behind the software. When I'm away from the keyboard, I'm usually watching movies or sports, solving puzzles, or hanging out with friends.
Python
Machine Learning
Data Analysis
Java
MySQL
MongoDB
C/C++
HTML/CSS
JavaScript
PHP
A research proposal for a custom lightweight CNN architecture designed to detect digital artifacts in deepfake media.
Deepfake media is rapidly destabilizing trust in social media, requiring lightweight detection tools.
Achieved 91% accuracy in artifact detection using a custom CNN model, outperforming similar lightweight architectures.
An AI-powered brain tumor analysis and clinical reporting platform that connects hospital administrators, radiologists, doctors, and patients through a centralized workflow.
Brain tumor assessment involves multiple stages of medical image analysis, radiologist review, doctor verification, reporting, and patient access, requiring a coordinated clinical workflow.
BrainInsight integrates AI-based tumor classification and segmentation with automated report generation, radiologist review, doctor verification, patient management, and hospital work assignment.
Creation of a usable, trusted, and ready-to-use dataset covering 40 distinct words of Bangla Sign Language to facilitate robust model training.
Existing datasets for Bangla Sign Language often lack diversity or reliability, hindering effective research.
Curated a clean, annotated dataset of 40 words, specifically processed to be "ready-to-use" for immediate training.
Selected experiments and academic works
Java-based application for managing bank transactions, utility transactions, and user records.
Check out my work and connect with me
Feel free to reach out ā I'll get back to you within 24 hours.
nadimdewan789@gmail.com
+880172117946
Dhaka, Bangladesh