Muntasir Mohammad
ML Engineer || Software Engineer || Data Scientist
About me
I am Muntasir Mohammad, Specialist, Data Science currently working at Robi Axiata Ltd. I graduated from North South University on 2020 on Computer Science and Engineering (major Machine Learning), securing the honour of Magna Cum Lauda with C.G.P.A - 3.75.
Before graduating, I did an internship at Gaze Technology, which then was an AI firm but now has pivoted to being a Software Firm. During that internship, I worked on Face Spoof Detection as one of their upcoming projects. After graduating, I started to work at NybSys from October 2020 and continued till July 2021. At NybSys, I worked with creating Machine Learning architectures for their upcoming projects, as well as working on their existing products. I had worked on NybFace, which is a Face Recognition based Access Control System. I had worked on developing the backend of NybFace for integrating with web and mobile based applications. I had also worked on creating a production build of Survello, which is a surveillance system. I had also contributed in performing OPS on the products, which included load testing the server, API testing, server hosting etc. We also worked on Big Data using AWS S3, AWS EMR, Apache Spark, Hadoop etc.
At Robi, I work on providing analytics on various existing user facing services of Robi and producing solutions to them. The solutions lie between finding data driven solutions to producing machine learning based services.
Skills
Artificial Intelligence
Artificial Intelligence skills
Big Data Framework
Big Data Technologies
Backend
Backend Technologies
Database
Database Technologies
Production Server
Production Server Deployment
Others
Other Valuable Skills
Experience
Gaze Technologies
NybSys Pvt. Ltd.
Robi Axiata Ltd
PROJECTS & PUBLICATIONS
BANNER
Banner: A Cost-Sensitive Contextualized Model for Bangla Named Entity Recognition
IEEE Access
Publication Date: 23 March, 2020
DOI: 10.1109/ACCESS.2020.2982427
FEW SHOT LEARNING
Attention Guided Relation Network for Few-Shot Image Classification
ICCCM 2019: The 7th International Conference on Computer and Communications ManagementAt: Bangkok, Thailand
Publication Date: July, 2019
DOI: 10.1145/3348445.3348481
Question Answer Generation
Bangla Question Answer Generation from Single Sentences
Performed Question Answer Generation downstream task on Bangla Language. Introduced Entity Prediction Network for predicting entities, Relation Extraction Network for predicting relation between predicted entities and Question Answer Generation Network for producing questions based on predicted entities & relation
SATGAN
SATGAN: Image Generation from Text Descriptions using Self Attention Generative Adversarial Network
Using 2 layer Generator and Discriminator Networks and Self Attention as attention mechanism, generated images with relatively good accuracy, comparable to state-of-the-art architectures
Binge Recommendation Engine
Binge Recommendation Engine
Built and deployed a recommendation engine for the Over The Top (OTT) platform Binge, a product of Robi. The recommendation engine comprises of a blend of content filtering and collaborative filtering recommendation engines. The recommendation engine gives 2x more accuracy in comparison to its predecessor.
NYBFACE
NybFace
NybFace: Face Recognition based Access Control System
NybFace is an Access Control System which leverages Face Recognition and Face Spoofing based technologies. I, along with my team, worked on developing the backend of NybFace for integrating with web and mobile based applications.
AFS Seair Freight Ltd
AFS Seair Freight Ltd
Built a responsive website for a freight forwarding company named ‘AFS Seair Freight Services’, https://afsbd.com. Used Bootstrap, Jquery as frameworks. Also designed the logo and business profile for the company using Adobe Illustrator. Currently working on creating an ERP based web application software to automate the internal workflow of the company. Used Stack (React, Flask, MongoDB, Node)
Deal Finder
Deal Finder
A unified platform for finding deals on different categories from various institutions through a web based application