Posts by Collection

National Girls’ Programming Contest

Became 10th in the National rounds, , 2021

Best Student Poster in 8th NSysS

Won the best student poster title, , 2021

• Worked on the posthumous brain of goats for signal processing and gaining valuable insights.

HerWill-NSDC DataForge

Became Champion, , 2023

• A Natural Language Processing task was given for the detection of fake news
• Processed text data with the help of lemmatization, tokenization, stemming etc.

BioMed Datathon

Became Runner-Up, kaggle, 2024

• A multilabel multiclass classification task of various heart-valve related diseases.
• Used a deep learning approach to solve the problem and achieved a good ICBHI score

ML Olympiad - CO2 Emissions Prediction Challenge

Secured a score higher than 30% of the contestants, kaggle, 2024

• Prediction of the amount of CO_2 by analyzing time-series data.
• Used the model LSTM for predicting missing values and ARIMA and LSTM for the final prediction.

SUST DLEnigma1.0

Became 8th out of 98 teams, kaggle, 2024

• A Deep Learning based competition on object detection specifically, vehicle detection in the context of the big cities of Bangladesh
• Used different YOLO and transformer-based models

C Compiler

GitHub Repository
Language: Flex, Bison, YACC

• Created a C compiler using tools like Flex, Bison, and Yacc, demonstrating my skills in buildingsoftware that understands and processes programming languages.

FireCop

Materials: Arduino UNO, Flame Sensor

• FireCop is a miniature firefighting robot built as a project under our Micro-processor and Micro-controller course.
• This robot was motivated with an objective of mitigating fire in a small space and with less cost.

4-bit MIPS Implementation

Materials: atmega32, basic ICs

• Designed a simplified 4-bit processor that implements a given reduced version of the MIPSinstruction set.
• Simulated the workflow in a preferred software (Logisim).
• Implemented the software design with hardware.

Online Exam Management System

GitHub Repository
Framework: React, NodeJS, PostgreSQL

• An online written and MCQ format exam management platform
• Two kinds of users- teachers and students.
• Teachers can monitor the performance of the students as individuals as well as a part of a targeted group with summary statistics.
• Students can keep track of their progress

Ray Tracing

GitHub Repository

Language: C++
Library: OpenGL

• Developed a ray tracing pipeline entirely from scratch, incorporating the Phong Lighting Model for accurate lighting effects and Recursive Reflection for realistic rendering.
• This pipeline enables the simulation of complex light interactions and creates visually stunning scenes with lifelike reflections and shading.

Sheba.xyz

GitHub Repository

Framework: ejs, NodeJS, OracleDB

• This is a simple replica of Sheba.xyz where a seller can add services and a customer can take the services needed.

Vehicle Detection in the Context of Bangladesh

GitHub Repository

Models used: YOLOv6, YOLOv8, Co-DETR, RTDETR

• Developed a pipeline for detecting vehicles in the context of Bangladesh using the BadODD dataset, which contains over 5,000 images captured across 9 cities in the country.
• Experimented with several pretrained models, including YOLO variants and detection transformers. RTDETR was selected as the most suitable model, as it effectively balances the strengths of both YOLO models and DETRs.

Is It Really Dead? Digging into Dead Brains through Analyzing Its Behavior in Response to Inducing External Impulses

Published in 8th NSysS, 2021

Supervisor: A. B. M. Alim Al Islam

• Even though live brains are widely studied in the literature, dead brains remain little explored as perhaps it is generally believed to have a dead brain not workable anymore.
• In contrary to such general perception, in this work, we explore the possibility of making a dead brain work again.
• Explored outputs from a dead brain in response to applying external stimuli (in the form of electrical signals) to it.

Poster Link

Download Paper

Comprehensive Feature Selection and Biological Relevance Analysis in Multi-Disease Gene Expression Data

Supervisor: Dr. Mohammad Saifur Rahman

• In this work, I have undertaken a comprehensive gene expression analysis of microarray cancer datasets, employing advanced techniques such as feature engineering and machine learning ensembles.
• By integrating SHAP (SHapley Additive exPlanations) and conducting enrichment analysis, I have provided a robust framework for interpreting the results, contributing valuable insights into the complex landscape of cancer genetics

Github repository

Download Paper

Protein Solubility Prediction

Supervisor: Dr. Yang Zhang

• Developed a deep learning-based pipeline to predict the solubility of protein sequences, utilizing a range of protein-related features.
• Incorporated advanced protein language models (LLMs) for feature extraction and sequence processing.

Summer 2023: Machine Learning Intern

Published: May 15, 2023

Duties included: Predicting the faulty behavior of the machines at a powerplant with necessary correlations to the features using basic Machine Learning algorithms, ensembles, and deep learning pipelines.

Summer 2024: Research Intern

Published: June 01, 2024

Duties include: Working with Enzyme and Protein characteristic prediction with Machine Learning and Deep Learning Pipelines
Supervisor: Dr. Yang Zhang

Adjunct Faculty

Published: October 15, 2024

Duties include: Working as a part-time faculty in the Department of Computer Science and Engineering.
Taught courses: Digital Logic Design, Artificial Intelligence and Machine Learning