I am a first-year PhD student in Computer Science at
University of Illinois Chicago. Currently, I am also working as a Teaching Assistant in Computer Science Department
at the University of Illinois Chicago. I have completed my B.Sc. in Computer Science and Engineering from
Bangladesh University of Engineering and Technology(BUET).
My research interests lie in the intersection of causal inference, data science, and machine learning. In particular, I am
interested in causal reasoning with data that can answer questions and/or solve problems in e-commerce, social media,
social science, and healthcare domains. I am fascinated in
applying econometrics methods for answering causal questions with data that can reason the past, and also provide insights
for the future. I am also interested in protecting user data privacy in software applications.
PhD Student in Computer Science(2023-Present)
University of Illinois Chicago
B.Sc in Computer Science and Engineering, 2018
Bangladesh University of Engineering and Technology(BUET), Dhaka, Bangladesh
Higher Secondary Certificate, 2013
Notre Dame College, Dhaka, Bangladesh
Keywords : Scene Reconstruction, Missing Scene Construction, Video Processing, Object Monitoring, Object Tracking, Surveillance System, Polyfit
We have introduced and developed a system that can present information of a human or an object to a user or visual inspector by analyzing videos taken from multiple sequential or partially overlapping static CCTV data from area of interests. It can reduce manual effort of visual inspector on visual surveillance system. For implementing the system, we need to recognize the input object in different video files taken from CCTV cameras. For recognizing the input object from frame to frame in video files, we have used Speeded Up Robust Features(SURF). The recognized object in a frame has been used as a new input for recognizing the input object in next video frame so that the input object can be detected more accurately using SURF as object shape or orientation can be changed over time. If there is a place in the area of interests where camera is defective or unavailable, we have developed a movement prediction algorithm for predicting the movement of the input object in that specific place using the objects previous or next velocities in different cameras. For determining the average velocity of the input object in a camera, we have considered the total changes of pixels over time and then converted the pixel distance to actual meter distance. As meter distance between two pixels is not constant in relative to the center of the camera, we have built a four degree polynomial approximation to determine meter distance from pixel distance. We have used these average velocities of the input object from different cameras to predict the movement of the object in the place of missing or defective camera on given the background information of that place.
Projects on which I have worked in Ridmik Labs :
Bangla article summarizer outputs an extractive Bangla summary of a given Bangla news article which is currently being used to generate automated Bangla summary of news articles for Ridmik News app
Boitoi is a JavaScript based secured Android e-book and short story reader with digital in-App book store. It is currently the most popular eBook reader in Bangladesh which has around 1M+ users.
Boitoi is a JavaScript based secured e-book reader for iOS with book store
Ridmik Keyboard is a Android keyboard in Bangladesh with Bangla phonetic keyboard, English, and Arabic layout which has 50M+ users
RidmikChat is an XMPP based secured messaging and social platform application for Android
User can login to any Ridmik app in andorid or iOS devices using a single Ridmik account through this library
AIST Project is a web application for users to visualize the predictions of AIST (An Interpretable Attention-based Deep Learning Model for Crime Prediction) model and analyze the predicted and actual crime data in different locations.
Users can report for crimes in police station, connect with nearby police station or friends in emergency, get notification alert, see verified crimes in an area and police can see reported crimes, connect with victims and verify crimes, make an area as danger prone.
A model is trained using LSTM on multiple attributes of share market data which can predict the starting price for next day share market.
Students can give exams and see marks on both MCQ and descriptive questions. Teachers can upload questions from admin panel, and also can check descriptive answers.
Teachers can upload question paper from admin panel. Students can give MCQ exam. and also can see their marks and all previous marks history on every subject.
Users can play this funny game in desktop and can see leaderboards.
Small scale C Compiler project which has four steps : symbol table, lexical analysis, parsing and code generation.
A computer graphics course assignment on which I have generated a realistic image of some common shape using OpenGL.
Smart sleeping mat is a hardware project where a user can set a time for alarm, and until a user goes out of the mat on the time of alarm, alarm won't stop. This will be helpful for those who accidentally close alarm during sleep. For the patient who needs to sleep in one particular direction, user can calibrate the mat according to that, and if the user moves aside from that particular direction, alarm will go on so that user can move to his desired position again.
User can play this game using android device accelerometer, see leaderboards and change theme for game screen. The speed of the game increases as time passes by. Game will be over if a user dies three times and user can restart the game again.
User can play this game in desktop, see leaderboards. The speed of the game increases as time passes by. Game will be over if a user dies three times.
Users can play and see leaderboard from android, iOS and desktop.
User can save, edit and delete one's writings in the notebook.
Users can see one's current velocity, current location in map and location details.
A user needs to answer some questions and based on his answers, his mental age will be generated.
Users can use this as a torch light from android device.
Programming Languages: Java, Python, Kotlin, Swift, C, C++, Assembly(80X86)
Scripting Languages: JavaScript, HTML, CSS, MATLAB, Shell Scripts
Mobile Application Development & Frameworks: Android, iOS, Django
Database: Oracle, SQL
Security: AES(CBC & ECB), CommpnCrypto API, iOS Keychain
Tools & Others : Pandas, NLTK, REST API, Beautiful Soup, Seaborn, Google ML Kit &
Vision API, Tensorflow, Google Map API, Firebase, JNI, JavaFX, LATEX, tkinter