EXTERNAL PROFILES
Assistant Professor
Md Iftekharul Alam Efat
I particularly enjoy collaborating with Software Systems with Artificial Intelligence and solve new challenges. Currently, doing research on "Precision Agriculture and Digital Twin Technologies", using machine learning approaches to enhance production of Agriculture with smart way.
Institute of Information Technology (IIT)
BIOGRAPHY
I am Md. Iftekharul Alam Efat, an Assistant Professor at the Institute of Information Technology (IIT), Noakhali Science and Technology University (NSTU). I hold both a BSc and an MSc in Software Engineering from the Institute of Information Technology (IIT), University of Dhaka (DU).I am an enthusiastic, adaptive, and fast learner with a strong passion for Secure Software Design & Architecture. I enjoy collaborating on software systems that integrate Artificial Intelligence and thrive on solving complex, real-world challenges. Recently, I’ve conducted research on IoT-Based Health Status Monitoring Technology, where I used neural networks to build a system capable of continuously monitoring the health conditions of diabetic patients. Currently, doing research on "Automated Bug Triaging and Classification using Machine Learning Techniques", applying state-of-the-art ML techniques on software repositories. My aim is to contribute to the development of intelligent and secure healthcare solutions through cutting-edge research and innovation.
RESEARCH INTERESTS
2013 - 2014
M.Sc. in Software Engineering
Institute of Information Technology (IIT)
University of Dhaka (DU), Bangladesh
Thesis: Thesis: Reusability Measurement for Software Components CGPA – 3.79 (in the scale of 4.00)
2009 - 2012
Bachelor of Information Technology
Major in Software Engineering
Institute of Information Technology (IIT), University of Dhaka (DU), Bangladesh
Thesis: Thesis: Face Recognition with Vibrant Local Ternary Pattern (VLTP) CGPA – 3.84 (in the scale of 4.00)
2006 - 2008
Higher Secondary Certificate
Science
Notre Dame College, Dhaka, Bangladesh
1996 - 2006
Secondary School Certificate
Science
St. Gregory’s High School, Dhaka, Bangladesh
Last updated on 2026-01-13 08:24:28
Sr. Lecturer
Software Engineering, Daffodil International University
2017-10-16 to
2019-02-12
2019-02-12
Lecturer
Software Engineering, Daffodil International University
2015-01-10 to
2017-10-15
2017-10-15
Product Manager
Solution Architecture Team, Four D Communicaitons Limited
2015-01-03 to
2017-04-30
2017-04-30
2014-04-26
Business Communication
Description: N/A
Organization: Project of Hi-Tech Park, Kaliakour, Ministry of ICT
Last Updated: 2025-10-07 13:29:45
2018-09-11
OBE Based Curriculum Design for Software Engineering Bachelor Program
Description: In the campus of Daffodil International University, during their Accreditation process.
Organization: ABET Accreditation, Thailand
Last Updated: 2025-10-07 13:48:48
AWARDS AND ACHIEVEMENTS
1
Best Presenter Award
ICIEV 2014
Description: 3rd International Conference on Informatics, Electronics & Vision (ICIEV '14) for paper: "Face Verification with Fully Dynamic Size Blocks based on Landmark Detection".2
University Undergraduate Scholarship of Excellence
University of Dhaka, Dhaka, Bangladesh
Description: Achieved scholarship for excellent result and other co-curricular activities throughout graduation period. Awarded by, University of Dhaka, Bangladesh3
2nd Runners-up
Banglalink Grandmaster Idea Contest
Description: Business plan and implementation of a VAS (Value Added Service) for Banglalink Idea: SMS based bus ticketing named EASY PASSANGER for inter-city & intra-city bus services in Bangladesh4
1st Runners-up on Project Showcasing
Islamic University of Technology (IUT), Gazipur, Bangladesh
Description: 3rd National ICT FEST 2011, Islamic University of Technology (IUT), Gazipur, Bangladesh Project: mobile based loan & payment, providing three features including account checking, mobile balance recharge, loan & payment using the Credit Rating idea.5
2nd Runners-up on Software Project Fair
East West University (EWU), Dhaka, Bangladesh
Description: EWU IT Festival 2011, East West University (EWU), Dhaka, Bangladesh Project: m-Tourism: A Mobile based Tourism System of seeking and booking6
Champion on BCSIR Science Fair
Bangladesh Council of Scientific and Industrial Research (BCSIR), Dhaka, Bangladesh
Description: Project: E-Voting: A Pathway to Free, Fair & Transparent Election7
Runners-up on 4th Divisional (Dhaka) Mathematics Olympiad
Bangladesh Mathematical Olympiad (BdMO)
2026
Journal
Lightweight temporal violence detection using YOLOv7-Tiny and Bi-GRU for resource-constrained environments
Authors: Hasanur Rahman Shishir, Nazmun Nahar, Mayeen Uddin Khandaker, Md Hasan Imam, Md. Iftekharul Alam Efat, Mohammad Kamrul Hasan
Journal: Systems and Soft Computing
DOI: https://doi.org/10.1016/j.sasc.2026.200494
2025
Journal
Blockchain Aided Smart Consensus Model for IoMT Architecture
Authors: Md. Iftekharul Alam Efat, Tasnim Rahman, Md Jane Alam, Shah Mostafa Khaled
Journal: Annals of Emerging Technologies in Computing (AETiC)
DOI: https://doi.org/10.33166/AETiC.2025.02.003
2022
Journal
A Systematic Review of Blockchain Applications
Authors: Farhana Akter Sunny, Petr Hajek, Michal Munk, Mohammad Zoynul Abedin, Md Shahriare Satu, Md Iftekharul Alam Efat, Md Jahidul Islam
Journal: IEEE Access
DOI: https://doi.org/10.1109/ACCESS.2022.3179690
2022
Journal
Identifying Optimised Speaker Identification Model using Hybrid GRU-CNN Feature Extraction Technique
Authors: Md. Iftekharul Alam Efat, Md Shazzad Hossain, Shuvra Aditya, Jahanggir Hossain Setu, KM Imtiaz-Ud-Din
Journal: International Journal of Computational Vision and Robotics
DOI: https://doi.org/10.1504/IJCVR.2022.126508
2022
Journal
Deep-learning Model using Hybrid Adaptive Trend Estimated Series for Modelling and Forecasting Sales
Authors: Md Iftekharul Alam Efat, Petr Hajek, Mohammad Zoynul Abedin, Rahat Uddin Azad, Md Al Jaber, Shuvra Aditya, Mohammad Kabir Hassan
Journal: Annals of Operations Research
DOI: https://doi.org/10.1007/s10479-022-04838-6
2020
Journal
Paperless Vehicle Certification (PVC) Framework for Transport Stakeholders
Authors: Md. Iftekharul Alam Efat, Shouvick Ahmed Naim, Tasnim Rahman
Journal: Journal of Noakhali Science and Technology University
2014
Journal
Dynamic Blocks for Face Verification
Authors: Mohammad Ibrahim, Tajkia Rahman Toma, Md Iftekharul Alam Efat, Shah Mostafa Khaled, Md Shariful Islam, Mohammad Shoyaib
Journal: International Journal of Computer Vision and Signal Processing
2013
Journal
Vision Inspired Local Ternary Pattern (VLTP) for Face Recognition & Verification
Authors: Md. Iftekharul Alam Efat, Mohammad Ibrahim, Humayun Kayesh, Shah Mostafa Khaled, Mohammad Mahbub Alam, Mohammad Shoyaib
Journal: Journal of Engineering and Technology
2024
Book Chapter
Book Name: Lecture Notes in Networks and Systems (LNNS)
Chapter Title: Optimized Energy Consumption in Healthcare WSNs Using Enhanced LEACH Protocol with K-Means
Authors: Md. Iftekharul Alam Efat, Tasnim Rahman, Shuvra Aditya, Kaushik Sarker
DOI: https://doi.org/10.1007/978-981-96-2721-9_21
2023
Book Chapter
Book Name: Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering (LNICST)
Chapter Title: Deep Convolutional Comparison Architecture for Breast Cancer Binary Classification
Authors: Nasim Ahmed Roni, Md Shazzad Hossain, Musarrat Bintay Hossain, Md Iftekharul Alam Efat, Mohammad Abu Yousuf
DOI: https://doi.org/10.1007/978-3-031-34619-4_16
2023
Book Chapter
Book Name: Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering (LNICST)
Chapter Title: Block-chain aided Cluster based Logistic Network for Food Supply Chain
Authors: Rahat Uddin Azad, Khair Ahammed, Muhammad Abdus Salam, Md. Ifthekarul Alam Efat
DOI: https://doi.org/10.1007/978-3-031-34622-4_34
2020
Book Chapter
Book Name: Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering (LNICST)
Chapter Title: IoT based Smart Health Monitoring System for Diabetes Patients using Neural Network
Authors: Md Iftekharul Alam Efat, Shoaib Rahman and Tasnim Rahman
DOI: https://doi.org/10.1007/978-3-030-52856-0_47
2021
Conference
A Hybrid GRU-CNN Feature Extraction Technique for Speaker Identification
Authors: Md. Shazzad Hossain Shihab, Shuvra Aditya, Jahangir Hossain Setu, KM Imtiaz-Ud-Din, Md. Iftekharul Alam Efat
DOI: https://doi.org/10.1109/ICCIT51783.2020.9392734
2020
Conference
Blockchain aided Vehicle Certification (BVC): A Secured E-Governance Framework for Transport Stakeholders
Authors: Moonmoon Das, Rahat Uddin Azad, Md Iftekharul Alam Efat
DOI: https://doi.org/10.1109/ICCIT51783.2020.9392725
2018
Conference
Trend Estimation of Stock Market: An Intelligent Decision System
Authors: Md. Iftekharul Alam Efat, Rakibul Bashar, KM Imtiaz-Ud-Din, Touhid Bhuiyan
2014
Conference
Dynamic Local Ternary Pattern for Face Recognition and Verification
Authors: Mohammad Ibrahim, Md Iftekharul Alam Efat, Humayun Kayesh Shamol, Shah Mostafa Khaled, Mohammad Shoyaib, M. Abdullah-Al-Wadud
DOI: ISBN: 978-960-474-361-2
2014
Conference
Face Verification with Fully Dynamic Size Blocks based on Landmark Detection
Authors: Mohammad Ibrahim, Md. Iftekharul Alam Efat, Shah Mostafa Khaled, Mohammad Shoyaib
DOI: https://doi.org/10.1109/ICIEV.2014.6850737
2014
Conference
Feature Prioritization for Analyzing and Enhancing Software Reusability
Authors: Md. Iftekharul Alam Efat, Md Saeed Siddik, Mohammad Shoyaib, Shah Mostafa Khaled
DOI: https://doi.org/10.1109/ICIEV.2014.6850738
2013
Conference
Automated Bangla Text Summarization by Sentence Scoring and Ranking
Authors: Md Iftekharul Alam Efat, Mohammad Ibrahim, Humayun Kayesh
DOI: https://doi.org/10.1109/ICIEV.2013.6572686
2026-07-11
to
Present
Research
Agri
Abstract: pricision agriculture
Last Updated: 2026-07-15 19:51:29
No Project Found
Last Updated: 2026-07-15 19:51:29
Course Content
👨🏫 Teaching Portfolio 📍 Noakhali Science and Technology University (2019 – Present) 🌱 Freshman 📐 Discrete Mathematics • 💻 Introduction to Software Engineering 🚀 Sophomore ⚙️ Algorithm Analysis & Design • 📋 Software Requirement Specification & Analysis • 🔐 Information Security 🏗️ Junior ⚖️ Professional Ethics for Software Engineering • 🧪 Software Testing & Quality Assurance • 🏛️ Software Design & Architecture • 🛡️ Software Security 🎯 Senior 📈 Software Project Management • 🔄 Software Maintenance 🎓 Graduate 📘 Formal Methods & Models in Software Engineering • 🤖 Soft Computing
No Course Materials Found....
Masters - 2021-22
Rahat Uddin Azad
Lecturer, Daffodil International University
Thesis Title: A Dual Encoder Fusion Model for Semantic-Aware Bangla Paraphrase Generation Using Bangla-T5 and Gemini Embedding
Overview: The primary objective of this research is to build and validate a paraphrase generation system for Bangla that maximizes semantic fidelity while producing meaningful, non-trivial re- expressions of the input sentences. A core challenge in achieving this goal is mitigating semantic drift, which can occur under conditions of limited and heterogeneous supervision. To address this issue, the methodology incorporates a stable, frozen semantic signal from Gemini Embeddings, which is integrated into both the representation and the objective functions of the model. This strategy is intended to ensure that the generated paraphrases retain semantic accuracy despite the inherent variability in the input data.
Honors - 2020-21
Tarekul Islam Tusher
Graduate
Thesis Title: Explainable Cross-Crop Generalization and Feature Bias in Deep Learning Models for Organic Crop Detection
Overview: Documentation-based verification of organic produce is not particularly popular, whereas nondestructive testing using lightweight convolutional neural networks (CNNs) has proven effective at distinguishing organic from non-organic apples, mushrooms, and cucumbers with more than 97% accuracy on RGB images. Whether it is learning a visual cue associated with organically grown food that abstracts beyond specific crops, or whether these classifiers are taking shortcuts in acquiring market data, remains uncertain. We train four classification pipelines (two with a fine-tuned MobileNetV2 and two with a fine-tuned EfficientNetB0) on a public RGB dataset containing images of organic apples, mushrooms, and cucumbers, and assess them using three different test schemes: train/validate/test, crop-wise 10-fold cross-validation and leave-one-crop-test generalization. With the conventional split, all four pipelines' test accuracy is higher than 97%, but with this baseline test required, the test accuracy drops to 50.8-57.9%, comparable to the majority-class baseline (50% balanced accuracy) for an unseen crop. A visualization of the model attention using Grad-CAM and a proposed but not yet released Produce Attention Score (PAS), which quantifies the portion of the model that is activated within the crop produce area after the area has been cropped, reveals that attention is only partly biased towards the crop produce (mean attention = 0.72 after cropping) and varies significantly between crop and between classes. The results as a whole reveal that, in this task, high in-domain test accuracy is not sufficient evidence to establish a transferable, crop-agnostic organic signal, and that cross-crop (or cross-domain) evaluation should be included as standard practice in attention-based auditing.
Honors - 2020-21
Mohammed Raju Ahmed
Graduate
Thesis Title: A Heterogeneous Deep Fusion Framework with SHAP-Based Explainability for AI-Generated Source Code Detection
Overview: As LLMs like ChatGPT become more popular for writing code from scratch, questions about academic honesty, source code provenance, and trustworthy software development are emerging. This paper introduces a novel hybrid approach for detecting sources in human-versus-AI (HVAI) source-code detection, which fuses three structurally different classifiers: XGBoost for representation based on TF-IDF and stylometric features, an LSTM network for tokenized code sequences, and a fine-tuned CodeBERT transformer for representation of context. Logistic Regression fusion layer combines the output probabilities of the three classifiers to get the final Human/AI decision. Experiments conducted in the HumanVSAI_CodeDataset (5,000 human-written, 5,000 AI-generated samples in Java, Python, C and C++) with random_state = 42, 80:20 split. The accuracy of XGBoost, LSTM and CodeBERT were 98.85%, 99.45% and 99.50% respectively on the held out test set, while the hybrid fusion model achieved 99.90% accuracy, precision, recall and f1 score with an ROC-AUC of 0.9998 on the held out test set. The same fusion model obtained 96.27% accuracy and 96.11% F1-score on the independent GPTSniffer RQ1-C1 Java benchmark (295 test samples), showing that performance can transfer from the in-distribution to other out-of-distribution datasets, but will not achieve the ceiling accuracy of the in-distribution. SHAP-based Explainability was used at the feature level and at the fusion level for the interpretation of the contribution of individual features and of each base model. An interface based on Streamlit was also created to showcase real-time probability estimation between the Human and AI. The results demonstrate measurable improvements using a combination of lexical, sequential, and contextual evidence over any single of the evidence types, as well as the limitations of generator diversity, stylometric-preprocessing reconstruction and the robustness to post-editing.
Honors - 2019-20
Imtiaz Chowdhury
Software Engineer
Thesis Title: A Framework for Maximizing Assertion Accuracy in LLMs Generated Test Code
Overview: Well designed and cost-effective unit tests are crucial to developing and maintaining quality software. Recently, Large Language Models have dramatically improved automated test generation; however, the precision of assertions in such tests is still an important issue to address. A novel framework is presented that uses a step-by-step approach to improve assertion reliability in LLM-generated test code by performing assertion quality classification, mutation-guided evaluation, and a targeted strategy for model fine-tuning. The framework was applied to a dataset of 1,328 test cases and shows substantial improvements in the accuracy of assertions. Empirical findings attest to the framework’s ability to enhance assertion power by at least double, while raising mutation scores by up to 93.57\%. The proposed framework addresses unique issues related to assertion generation and offers a highly effective approach to improving the dependability and efficiency of LLM-generated tests in actual software development scenarios.
Honors - 2019-20
Shoriful Habib
Software Engineer
Thesis Title: AI Driven Dynamic Multi Period Vehicle Routing
Overview: This thesis addresses the Dynamic Multi-Period Vehicle Routing Problem (DMPVRP) in order delivery logistics, focusing on efficient route planning for multiple vehicles under dynamic customer demands and operational constraints. A novel Graph Neural Network (GNN)-based framework is proposed to learn routing policies from heuristic-generated data and adaptively optimize delivery routes over multiple days. The GNN approach is evaluated against a classical Nearest Neighbor heuristic, demonstrating promising adaptability despite initial lower route efficiency. Experimental results show the heuristic achieves shorter total distance and lower fuel cost, while the GNN model offers scalability and potential for improvement through advanced training. This research bridges traditional optimization and modern AI, providing a foundation for scalable, robust dynamic routing solutions in real-world logistics.ew
Honors - 2017-18
Shuvra Aditya
MSc. Student, IIT, NSTU
Thesis Title: Explainable Artificial Intelligence (XAI) Enabled Machine Learning-based Intrusion Detection System to Enhance Interpretability and Trustworthiness
Overview: Despite the increasing popularity of machine learning models in cyber-security applications such as intrusion detection systems (IDS), most of these models are often perceived as black-box solutions. The significance of eXplainable Artificial Intelligence (XAI) has grown as it allows for the interpretation of machine learning models, thereby enhancing trust management. XAI enables human experts to comprehend the underlying data evidence and causal reasoning, leading to improved trust in the system. While previous studies have primarily focused on the accuracy of various classification algorithms for trust in IDS, they often lack insights into the behavior and reasoning of these complex algorithms. In this paper, we address the concept of XAI to enhance trust management in IDS by examining the decision tree model. By utilizing simple decision tree algorithms that are easily interpretable and resemble human decision-making processes, we split choices into smaller subchoices for IDS. We conducted experiments using a widely used KDD benchmark dataset and extracted rules to assess the effectiveness of this approach. Additionally, we compared the accuracy of the decision tree approach with other state-of-the-art algorithms.
- Institutional Email: iftekhar.iit@nstu.edu.bd
- Personal Email: iftekhar.efat@gmail.com
- Mobile number: 01727208714
- Emergency Contact: 01842208714
- PABX:1202
- Website:https://iftekhar.sererl.com/
SOCIAL PROFILES
Institute
Institute of Information Technology (IIT)
Noakhali Science and Technology University