Research Projects
Boa Python
Boa is a domain-specific language developed to facilitate mining software repositories (MSR) for research. It leverages the MapReduce algorithm to scale the mining process to hundreds of thousands of projects. Originally designed for Java-based projects, Boa eases MSR tasks significantly.
In this project, I extended Boa to support Python-based projects, enabling large-scale program analysis for the rapidly growing field of Python, particularly in machine learning. This involved adapting the Eclipse Dynamic Language Toolkit (DLTK) for Python, which had long been unsupported. I extended the DLTK Python to accommodate new features introduced in Python 3.0 and beyond. Additionally, I added support for converting parsed Python sources into a language-independent Abstract Syntax Tree (AST) format for Boa and implemented advanced program analysis features such as alias analysis and def-use chain analysis for Python.
Uncertainty Quantification for Deep Neural Network
This research-based project was developed as part of a graduate-level course at Iowa State University. In this project, we introduce a novel probabilistic first-order type system, Uncertain
High-Resolution Traffic Data Analysis System
During my time at Mississippi State University, I developed a comprehensive system for monitoring and analyzing the performance of the traffic signal system for the Mississippi Department of Transportation. This system provides an effective interface to analyze traffic data collected through Wi-Fi sensors placed at various road intersections. Notably, it enables the evaluation of traffic progressions on arterials using both the existing Purdue Coordination Diagram and a novel Multi-intersection Coordination Diagram. The system also forms the basis for addressing many other real-world traffic signal challenges, facilitated by Dr. Pengfei Li’s research group at the University of Texas at Arlington.
Industrial Projects
Prison Inmate Database System
As part of my role as a software engineer at TigerIT Bangladesh Ltd., I worked on a number of industrial-scale projects, including the prison inmate system. In this project, we developed a comprehensive system for efficiently indexing prisoners, overseen by the Rapid Action Battalion (RAB), a law-enforcing agency in Bangladesh. The project has three distinct modules: a .NET desktop module to enroll prisoners, including biometric information; a backend that serves as a Spring Web Service operating with the SOAP protocol; and an administrative web portal that provides efficient searching and configuration of the system.
I was a lead developer in this project, primarily responsible for the administrative portal and backend web services. To develop this system, we leveraged frameworks such as Spring and Hibernate. The administrative portal was developed using an MVC pattern enabled by Spring MVC. Security in both the web service and the portal was tightened through Spring Security, and the front end was designed using Bootstrap and JavaScript.
Machine Readable Passport System for Nepal Government
Similar to the Prison Inmate System, this project was also accomplished during my time at TigerIT Bangladesh Ltd. It had three distinct modules: a .NET desktop client, an administrative portal, and a backend web service. I was primarily involved in the development of the desktop client responsible for capturing the demographic and biometric information of users. To that end, I implemented this module using frameworks such as .NET and NHibernate. I also worked with the APIs of various biometric devices, including fingerprint scanners, iris scanners, document scanners, and signature pads, to integrate them with our enrollment system.
Academic Projects
Delta Debugging
As part of a class project at Iowa State University, I developed a tool that utilizes delta debugging to identify bug-inducing changes in C programs. This tool takes a set of changes and a test suite then employs a divide-and-conquer strategy to pinpoint the minimal set of changes that may have introduced a bug. The tool was implemented in Python.
Hydrogen Web
This project provides a web wrapper for a bug fixing patch verification tool called Hydrogen. Given a patch, it aims to identify whether it breaks existing functionality. I developed this web wrapper as part of a class project at Iowa State University. Given multiple versions of a file, it allows users to select which pairs to analyze and run analysis on a chosen pair of files by creating a Hydrogen Docker instance on-demand. It then allows users to visualize the analysis results interactively and produce relevant demographics for the patch. I used Spring Boot to develop the back end and spotify docker-client to manage on-demand Docker instances.
Demonstrative Projects
News App
This is a simple web application based on the Java Spring Boot framework for creating news stories and accessing them in different data formats for consumption. I developed this web application as part of an assessment for a Java Developer position at Cefalo in 2016, which led to a job offer from them. The application uses an in-memory SQL database for persisting data and Thymeleaf for rendering the HTML. The application has been developed using an MVC architectural pattern, promoting a loosely coupled design of the different components, making it scalable and easily maintainable. This application can serve as a good starting point for aspiring full-stack web developers aiming to leverage an MVC architecture and the Java ecosystem.
News App Again
This is the News App redone using a different technology stack as a single-page application (SPA). It uses a NoSQL database instead of an SQL one for its persistence system. The front end is built with AngularJS to render the view and design it as a SPA. Additionally, it automates both back-end and front-end dependency management using Maven and Bower.