Md Ataullha Saim

News & Updates
Md Ataullha Saim

I am an undergraduate student in Computer Science and Engineering at Shahjalal University of Science and Technology (SUST), Bangladesh, specializing in Speech Processing and Predictive Analytics under the guidance of Professor M. Shahidur Rahman, PhD.

Currently, I serve as an AI Engineer at Synesis IT, Convay. My research interests lie in Low-Resource Speech Processing, Natural Language Processing, and Time Series Analysis, with publications in esteemed journals such as Elsevier and IEEE.

For inquiries, email me at: ataullha00 AT gmail DOT com

Connect with me on: LinkedIn, GitHub, Google Scholar, ResearchGate, Kaggle, Curriculum Vitae (CV).

Professional Experience

Current Role

Artificial Intelligence Engineer - Synesis IT PLC, Dhaka, Bangladesh, Sep 2024 - Present

Specializing in creating sophisticated chatbots tailored to a wide array of applications, showcasing expertise in designing conversational AI systems that seamlessly integrate with business needs.

Work History

Natural Language Processing Engineer - Synesis IT PLC, Dhaka, Bangladesh, Mar 2024 - Aug 2024

Contributed to the enhancement of SIT-ASR, incorporating machine learning-driven spell suggestions to handle out-of-vocabulary (OOV) challenges. Developed a Bangla speaker diarization system, enabling accurate segmentation of speakers in audio. Built a custom LLM-based chatbot for Q&A applications and implemented English ASR support within SIT-ASR. Generated SIT-Live-CC for Bangla and English and refined the SIT Convay denoiser to elevate its performance and user experience.

Education

Shahjalal University of Science and Technology - Bachelor of Engineering (BE) in Computer Science, Sylhet, Bangladesh, Feb 2019 - Dec 2024

Achieved a CGPA of 3.74/4.00 while actively participating as a member of the CSE Society, SUST. During my final 1.5 academic years, I experienced significant academic growth, improving my CGPA by nearly 0.3, deepening my understanding of Machine Learning, and contributing to research with more than two publications. This period of transformation was profoundly influenced by the guidance and mentorship of Prof. Dr. M. Shahidur Rahman, for whom I am immensely grateful. My journey at SUST has equipped me with advanced skills in research, technical writing, project management, and programming, particularly in Python, along with expertise in Machine Learning.

Notre Dame College - Higher Secondary Certificate (HSC) in Science, Dhaka, Bangladesh, Jul 2016 - Jun 2018

Finished with a perfect GPA of 5.00/5.00, earning a General Grade Government Scholarship for academic excellence. During my time at Notre Dame College, I was involved in the Science and English Clubs, fostering both my analytical and communication skills. This period remains one of the most rewarding academic phases of my life, underscoring the importance of discipline and time management in achieving excellence.

Shamsul Hoque Khan School & College - Secondary School Certificate (SSC) in Science, Dhaka, Bangladesh, Jan 2009 - Feb 2016

Completed secondary education with a perfect GPA of 5.00/5.00. My time here laid the groundwork for my academic success, building a strong foundation in the sciences and fostering a lifelong commitment to learning and personal growth.

Research

Highlights

My research focuses on innovative applications of Artificial Intelligence in low-resource environments, with an emphasis on Large Language Models (LLMs), Bangla speaker diarization, Bangla ASR, and machine-learning-driven product. My goal is to enhance speech processing, natural language understanding, and text-to-speech (TTS) systems to create accessible AI solutions that drive progress in technology, education, and conservation.

Publications

Md Shahidul Islam, Pabel Shahrear, Goutam Saha, Md Ataullha, and M. Shahidur Rahman. "Mathematical analysis and prediction of future outbreak of dengue on time-varying contact rate using machine learning approach." Computers in Biology and Medicine, vol. 178, 2024, p. 108707. [Journal]

Md Ataullha, Mushfiqur Rahman, and M. Shahidur Rahman. "PakhiderChobi: A Comprehensive Dataset for Real-time Detection of Bangladeshi Birds." 2023 26th International Conference on Computer and Information Technology (ICCIT). IEEE, 2023. [Conference]

Md Ataullha, Soumik Paul Jisun, Ishrat Jahan, Mushfiqur Rahman. "BanFish: A Dataset for Classifying Common Bangladeshi Fish Species." 2024 6th International Conference on Electrical Engineering and Information & Communication Technology (ICEEICT). IEEE, 2024. [Conference]

Md Ataullha, Sumitra Das, and M. Shahidur Rahman. "Air Quality Index Analysis and Forecasting in Dhaka with ARIMA, LSTM, and Prophet." 2024 27th International Conference on Computer and Information Technology (ICCIT). IEEE, 2024.

Md Ataullha, Mahedi Hassan Rabby, Mushfiqur Rahman, Tahsina Bintay Azam "Bengali Document Layout Analysis with Detectron2." arXiv preprint arXiv:2308.13769, 2023. [arXiv]

Featured Projects

StarGaze BD - ML Algorithms (RF, SVM, LR), CNN, Flask, JavaScript, Streamlit

A machine learning and deep learning-based web application designed to identify prominent Bangladeshi personalities from photos. It utilizes custom-built ML algorithms (Random Forest, SVM, Logistic Regression) and a CNN model for high accuracy. Integrated using Flask for backend processing, and a user-friendly interface is implemented with JavaScript and Streamlit. [GitHub Link]

Freshman Utilities - Java, Firebase, Google Directions API, Jsoup

An Android application crafted to simplify campus life for university freshmen by offering features such as navigation through the Google Directions API, Firebase-based data storage for sharing resources, and web scraping with Jsoup to collect essential campus-related information. [GitHub Link]

CourseTrackr - Java, Servlet, JSP, MySQL

A comprehensive web-based course management system tailored for educational institutions. It features role-specific functionalities for admins, teachers, and students. Built using Java Servlets and JSP for dynamic web pages and MySQL for robust backend database management. [GitHub Link]

QuesTek Pro - Java, JavaFX, Text files

A JavaFX-based desktop application for conducting and managing multiple-choice quizzes. It supports admin verification, quiz result exportation, and local data storage with text files. Ideal for in-person examination setups. [GitHub Link]

Recent Talks & Interviews

Presentation at ICCIT'23 - 26th ICCIT, Cox's Bazar, December 2023

Delivered a presentation on the PakhiderChobi dataset, which consists of 8,670 annotated images of 33 Bangladeshi bird species. The talk highlighted the importance of accurate bird detection and classification for urban conservation efforts. Utilizing the YOLOv8 object detection model, the research achieved a mean average precision (mAP) of 95.3% with an inference time of just 6.6 milliseconds per image. This tool aims to support real-time recognition, even in challenging scenarios, and contributes to bird conservation efforts amidst urbanization and habitat loss. Watch the talk on [YouTube].

Contact

I’m always open to discussing new projects, research ideas, or collaboration opportunities. Please feel free to reach out by completing the contact form, and I’ll get back to you as soon as possible.

The form includes fields for your email, name, reference, address, phone number, and message. If you're requesting a meeting, you can also specify a preferred date and time for a 30-minute session. Please indicate the priority of your message on a scale from 1 (Low Priority) to 3 (High Priority). For urgent matters, kindly highlight this in your message so I can prioritize my response.