Mobasshir Bhuiya Shagor

Hi, I am Mobasshir

A passionate AI/ML Researcher and Engineer with extensive experience in Deep Learning, Natural Language Processing, Computer Vision, Generative AI, and Recommender Systems. I currently serve as the Lead Data/Machine Learning Engineer at Graaho, where I focus on AI-powered product development. Previously, I spent three years as a Data Scientist at Graaho and one and a half years as the Machine Learning Lead at Renforce, honing my Data Science, Machine Learning, and MLOps skills.

I hold a Bachelor's degree in Computer Science & Engineering from the American International University Bangladesh (AIUB) and a specialized training in Artificial Intelligence & Prototype Development from the National University of Singapore (NUS). My research interests encompass a diverse array of cutting-edge areas, including the development of advanced cost functions and optimization strategies for machine learning models, innovative evaluation metrics tailored for complex datasets, and the integration of Privacy and Explainability in AI. I am also deeply invested in exploring secure and efficient Edge AI technologies that prioritize both data privacy and model transparency. Beyond my professional pursuits, I enjoy exploring new places and have recently developed a passion for cooking delicious foods.

ML Resume

SE Resume

Whats New

Research

This project aims to revolutionize agriculture by integrating IoT, AI, and GIS technologies into traditional farming machinery. Autonomous robot tractors, enhanced with GPS and terrain mapping, will automate soil preparation, planting, and harvesting, while AI-driven mechanical claws enable precise crop collection. Smart drones equipped with computer vision will monitor crops, detect anomalies, and perform tasks like pesticide application. IoT-based sensors will monitor environmental conditions, automating irrigation and pest control. All systems are remotely operable and monitored, optimizing efficiency and reducing operational costs.

Smart Mechanization

Our team has created and generously open-sourced a comprehensive cow image dataset comprising 17,899 images, enriched with vital information including sex, color, breed, feed, age, teeth, height, weight, price, and size. This dataset is versatile, serving as a valuable resource for both classification and regression tasks. Additionally, we have contributed baseline models that showcase the dataset's utility for regression and classification purposes, all built using the Keras functional API, further advancing research in this domain.

Cow weight and Breed Prediction; CID

In our AWS project, we participated in SemEval 2022 Task 11, specifically focusing on MultiCoNER Multilingual Complex Named Entity Recognition. Our research involved extending the base model and achieving substantial improvements in precision, recall, and F1 score values. This enhancement was accomplished through a strategic modification, replacing the CRF layer with a more effective BiLSTM-CRF layer, demonstrating our commitment to advancing the state-of-the-art in NER techniques within multilingual contexts.

MultiCoNER

Our research focused on advancing recommendation systems beyond traditional approaches. We developed KRE, a system capable of capturing contextual information such as weather data, day of the week, and time of day to enhance recommendations. We initially explored machine learning-based methods using libraries like Surprise and Implicit, then transitioned to DLRM, which effectively addressed cold start and sparse data challenges. Currently, we are dedicated to optimizing our model for real-time, low-latency online predictions. This research was conducted under the guidance of NUS professors as part of their prototype development program, highlighting our commitment to innovative and practical solutions in recommendation technology.

KhaoDao Recommendation Engine

Our research centered on the development of a computer vision application employing instance segmentation techniques, specifically Mask R-CNN (MRCNN), to extract precise coordinates related to the cow's girth, front, and back. Subsequently, we designed a multi-input-output network architecture using Keras functional API to predict essential parameters such as height, length, weight, and breed. To address class imbalance challenges in length and weight prediction, we adopted a regression-based approach, underscoring our commitment to innovative problem-solving in the field of computer vision and agriculture.

Smart Guess - Bengal Meat

Our research uses human affective responses to classify places based on problem categories, enhancing our understanding of human interactions with environments and supporting smart geospatial applications. We introduce the Categorized Affect Map (CAM) and a category-affect-space model, redefining crowdsourced data collection methods. Our study contributes to GIScience by improving our understanding of human behavior in diverse locations. We employ a multi-label classification algorithm to categorize places and address data collection challenges in affective response crowdsourcing.

Categorized Affect Map

Publications

ICCA ’22: Proceedings of the 2nd International Conference on Computing Advancements - March 2022

Mobasshir Bhuiya Shagor, Md Zahidul Haque Alvi, Khandaker Tabin Hasan

"CID: Cow Images Dataset for Regression and Classification"

ICCA 2020: Proceedings of the International Conference on Computing Advancements - January 2020

Mobasshir Bhuiya Shagor, Mahmood Al Rayhan, Kazi Imdadul Huq, Irin Nahar, Khandaker Tabin Hasan

"CAM - Categorized Affect Map: Implementation of a Smart Geographic Information System for Categorizing Places Based on People’s Affective Responses"

Honors & Awards

I'm honored to be recognized as one of the top 5% of data scientists on Kaggle, demonstrating my proficiency in data analysis and machine learning. My expertise extends to being a trusted expert in creating informative notebooks, curating datasets, and actively engaging in insightful discussions within the Kaggle community. I am committed to pushing the boundaries of data science and fostering knowledge sharing in this dynamic field.

Kaggle 3x Expert - Apr 2021

In our pursuit of addressing challenges related to Sustainable Development Goals, we introduced the Categorized Affect Map, a powerful visualization tool that offers deeper insights into real-world contexts. Our team's dedication was recognized with a prestigious top 10 position, underscoring the value of our work. We look forward to collaborating with A2i to further refine and implement this innovative concept for more accurate Minimum Viable Product (MVP) development in this field.

Top 10 innovative teams in Sustainable Development Goal Hackathon 2018, Banglalink Digital & A2i - Dec 2018

I proudly achieved the position of 2nd runner-up in the AI, Agriculture category with "Smart Guess," a cutting-edge computer vision application. This accomplishment reflects my dedication and expertise in leveraging AI for agricultural advancements, contributing to more efficient and innovative farming practices. I am committed to continuing my work in the intersection of AI and agriculture to drive positive impacts in the field.

Basis National ICT Award 2019 - Oct 2019

During my tenure in the Specialist Training on Artificial Intelligence & Prototype Development Programme for IT/ITES Industry Leaders, organized by the Singapore e-Government Leadership Center and the Ministry of ICT in Bangladesh, I gained invaluable insights into cutting-edge AI technologies and prototype development. This experience allowed me to enhance my skills and knowledge, positioning me as a leader in the IT/ITES industry. I am now better equipped to drive innovation and contribute to the advancement of technology in my field.

Specialist Training on Artificial Intelligence & Prototype Development Programme for IT/ITES Industry Leaders - Jun 2020

I dedicated four months of rigorous effort to lead and coordinate various activities, including meetups, webinars, quizzes, study jams, and team collaboration, as part of the Bertelsmann Cloud Track Program. My commitment and impact were acknowledged with the "You Rock" award in the #JackOfAllTrades category, a recognition that was based on active votes from a community of approximately 5,000 international students. This experience, complete with an official site feature, certificate, and valuable prizes, served as a truly inspiring milestone in my professional journey.

Jack Of All Trades [You Rock Award] - Mar 2020

I dedicated four months of relentless effort to organize meetups, webinars, quizzes, study jams, and foster a supportive community spirit among my peers in the Intel Edge AI Scholarship Program. As a result of my contributions, I was honored with the prestigious "You Rock" award in the #Kahuna category, a recognition bestowed based on active votes from over 13,000 international students. This achievement, which includes being featured on their official site and receiving a certificate and valuable prizes, has been a truly inspiring milestone in my journey.

Kahuna [You Rock Award] - Mar 2020

I am thrilled to have been appointed as the Machine Learning Lead, entrusted with the pivotal responsibility of establishing a robust ecosystem for Big Data and Machine Learning. In this role, I will be leveraging my expertise to design and implement cutting-edge solutions that empower data-driven decision-making and innovation within the organization. I am excited to drive forward this transformative initiative and contribute to the advancement of data science and machine learning.

Machine Learning Lead, Renforce, Australia - Jan 2021

Receiving the junior scholarship at the general grade level was a pivotal achievement in my life, marking a significant milestone in my journey towards a prosperous future. This accomplishment not only recognizes my dedication and hard work but also serves as a gateway to numerous opportunities and a brighter tomorrow. I am determined to continue striving for excellence and making the most of this scholarship's potential for personal and professional growth.

Junior Scholarship, Intermediate and Secondary Education Boards, Bangladesh - Mar 2009

Page updated

Google Sites

Report abuse