Game Theory & Reinforcement Learning
I am studying papers and books on Reinforcement Learning and implementing in python and Tensorflow as a part of the
Reinforcement Learning course.
Fall 2019, AI in Games.
AI in Games produces new challenges for AI researchers. So, I am pursuing studies in this domain to discover core problems to solve in AI.
Fall 2019, Data Mining.
Taking datamining course to brush up some forgotten skills.
Spring 2019, Machine Learning.
A more recent course to learn deep learning advancements.
Fall, 2018, Spark Streaming & Spark SQL
I studied on spark streaming and underlying architecture to implement a transformation or
application in it as a part of the course - Design and Implementation of Database Systems. Spark & Scala.
Previously, I thought I knew a lot about production data management challenges and technologies.
But taking a course with Professor Sheldon Finkelstein bewildered and amused me – it is a whole
new level. I got my chance to learn and question the internal architectures of industry leading
database systems and got introduced to practical and theoretical problems that I have never
imagined before. For example, I learned that with effort on optimizing communication patterns, we can
create solutions which stands between Shared-Nothing and Shared-Everything architectures and gain better
performance in practical scenarios.
Fall, 2018, Supervised Learning
I studied statistical machine learning and deep learning as a part of the course – Data Mining and Business Analytics.
It also included, the Doodle Challenge, a competition on Kaggle. R, Python, & Keras.
Data Mining and Business Analytics course with Ram Akella helped me to think about read-world data in different aspects,
and what exactly different models and their configuration elements mean, and which assumptions they have made.
Summar, 2019, Distributed Computing Intern
Futurewei Technologies Ltd
Resouce Scheduling with Machine Learning
This was the main goal. Schedule resources in cloud and edge networks with Machine Learning tools. I offered to research on reinforcment
approaches to that networks can learn to adapt. Unfortunately, they shut down before I could finish my plan.
Kubernetes and Test bed
We created the test bed with Kubernetes. I fixed Kafka library for node.js.
2017 – 2018, Lead of Research-Based Projects
In association with Reve Systems Ltd, BUET & ICT Ministry
Old is gold. It has its idiosyncrasies and challenges. I read about 50 recent papers on
page/layout segmentation, regions detection, noise deduction, and neural approaches. If you are interested check out
ICDAR 2017 papers.
. My main focus was layout analysis and page segmentation.
It is a good place to both start and get into the depths of OCR.
Bengali Grammar and Spelling Checker
Tools for automatic grammar and spell checking in Bengali. Researching n-gram and recurrent neural network (RNN)
algorithms. I have developed a fast and efficient algorithm for suggesting correct words in case of
replacement error in words recently. It is faster than edit distance.
Bengali Sign Language Translator
Sign language to text, text to animated sign language for Bengali language. Prediction of hand skeleton from static image,
combining outcome of multiple images to predict movement of hand, make system adaptive to different camera view, train the data
collected from hand movement and predict the gesture. We have used Support Vector Machine for the android implementation, and
Convolutional Neural Network for python-based desktop implementation.
My next challenge is to research and improve dynamic gesture recognition.
Bengali Unicode to IPA font transcription
Currently developing a rule-based algorithm to automatically transcribe Bengali text to IPA phonetics
language for voice synthesizers.
We will work on contextual variations of Bengali pronunciation next.
2016 – 2018, Senior Project Manager and Software Architect
Reve Systems Ltd
Language Training Application
Developed real time communication models for the virtual class-room.
It has been recognized as the best project of the ICT ministry of 2017 and will be used to
train millions of people in Bangladesh.
Link to Vashaguru
Law and Case Search Engine
Mongo and Apache Solr based search engine. Written in Spring Boot.
Disaster Management Platform, Armed Forces Division, Bangladesh
MySQL, MongoDB, Servlet & Struts.
2010 – 2015, CTO & CEO
ICode Bangladesh, Bangladesh
We started as a software outsourcing company. Inistially, I developed all the systems from scratch.
I ended up managing a team of 30 before we shut it down.
2006 - 2010
Freelance Web Developer
I became and elite developer in the most popular freelance platform, freelancer.com, in 2009.
I mostly worked in LAMP platform.
I am into a lot of things. I am trying to cut back on my hobbies. Right now I am holding on to
Music, Cooking, Table Tennis, Jogging, Potlucks, Couchsurfing Meet-ups, and any event with FREE food and drinks.
I am looking forward to learning surfing. If you can teach me in exchange for my wild recipes, do let me know.
Recently started making instrumental tracks on my electric guitar. Mostly experimental with Blues and Classical influence.