Salauddin Tapu

Junior Machine Learning Engineer | Anwar Technologies

Dhaka, Bangladesh

About Me

Hello! I am Salauddin Tapu, an AI enthusiast with expertise in machine learning, deep learning, and computer vision. The timeless question, "Can machines think?", has sparked my curiosity about the transformative capabilities of AI, which gradually evolved into a focused passion for building intelligent systems that can make a real-world impact. During my undergraduate years, this passion led me to explore how AI can contribute to humankind, and since then, I’ve been fortunate to work on projects that bring cutting-edge research to life. Over the years, I’ve had the privilege of working on a range of impactful, real-world AI projects—building robust systems that bridge theory and application. My work spans from intelligent document analysis to real-time facial recognition, always driven by a desire to solve meaningful problems. Fueled by an insatiable hunger for problem-solving and a passion for intelligent systems, I aspire to tackle real-world challenges through cutting-edge AI research that creates a meaningful, positive impact for humanity.

Academic Background

Khulna University

2018 - 2023 | B.Sc | Electronics & Communication Engineering Discipline (Supervisor: Dr. Uzzal Biswas)

Thesis: Prediction of Arrhythmia after Acute Myocardial Infarction Using Machine Learning and Statistical Techniques

Research Interests

Machine Learning
Deep Learning
Computer Vision (Image Segmentation, Object Detection & Tracking, Biomedical Image Processing)
Large Language Models
Natural Language Processing
Multimodal Learning

Publications

Chowdhury, M. S., Tapu, S., Sarkar, N. K., Ali, F. B., & Sabrin, L. (2025). Med-2D SegNet: A Light Weight Deep Neural Network for Medical 2D Image Segmentation. arXiv preprint arXiv:2504.14715. [Paper]
Prediction of Arrhythmia after Acute Myocardial Infarction Using Machine Learning. [Submitted - Under Review]
Gronthy, U. U., Biswas, U., Tapu, S., Samad, M. A., & Nahid, A. A. (2023). A bibliometric analysis on arrhythmia detection and classification from 2005 to 2022. Diagnostics, 13(10), 1732. [Paper]

Research Experience

Prediction of Arrhythmia after Acute Myocardial Infarction Using Machine Learning and Statistical Techniques

Developed a machine learning framework for predicting arrhythmias after acute myocardial infarction (AMI), leveraging Random Forest and the Cuckoo Search Algorithm (meta-heuristic) for feature selection and hyperparameter tuning. Identified key predictive features and visualized data distribution using t-SNE, achieving improved classification performance.

Med-2D SegNet: A Light Weight Deep Neural Network for Medical 2D Image Segmentation

Developed Med-2D SegNet, a lightweight segmentation model achieving 89.77% Dice similarity across 20 datasets from 12 medical imaging modalities. Utilizing only 2.07 million parameters, its innovative Med Block optimizes dimensionality reduction and feature extraction, excelling in zero-shot learning and multi-domain tasks.

Work Experience

Anwar Technologies

Nov 2023 - Present | Junior Machine Learning Engineer (Full Time)

Aug 2023 - Nov 2023 | Machine Learning Engineer (Intern)

VISIE Limited

Jan 2023 - Dec 2024 | Junior Machine Learning Engineer (Project-Based)

Projects

Bangla Handwritten OCR

Bangla Text Recognition

Med-2D SegNet

Spider plot comparing the performance of the proposed method with state-of-the-art works, highlighting its superior and comparable Dice score and exceptional performance across various evaluation metrics.

Comparison of model parameters vs. DSC on the KVASIR-SEG dataset. Our proposed method achieved a 95.78% Dice score, outperforming competing models with the lowest parameter count, demonstrating an optimal balance between efficiency and segmentation accuracy.