Mohammad Rafieian

Mohammad Rafieian

Ph.D. Candidate · Software Engineering & Security

University of Texas at Dallas · Advised by Dr. Shiyi Wei

Program Analysis Software Security Automated Testing Machine Learning
View Research Get In Touch

01 About Me

I am a Ph.D. candidate in Software Engineering at the University of Texas at Dallas, advised by Dr. Shiyi Wei. My research lies at the intersection of program analysis, software security, and machine learning — building automated techniques that make static analysis and testing tools more sound, precise, and trustworthy.

I received my M.S. in May 2026 and expect to complete my Ph.D. in May 2027. I am open to research-oriented and software engineering roles in both industry and academia.

Education

Ph.D., Software Engineering & Security

University of Texas at Dallas

2023 – Present · Expected May 2027 | GPA: 3.8

Advisor: Dr. Shiyi Wei

M.S., Software Engineering & Security

University of Texas at Dallas

Conferred May 2026

B.S., Computer Engineering

Isfahan University

2019 – 2023

Skills

Languages

Python Java C JavaScript Scala SQL

Machine Learning

PyTorch TensorFlow Scikit-learn finetuning LLMs agentic systems

Program Analysis

WALA DOOP Soot jazzer pointer analysis call graph analysis Opal

Frameworks

Django Flask Node.js React.js ASM

Development Tools

Git Docker Linux PostgreSQL Redis

02 Research & Publications

Research Interests

🧠

Machine Learning for Program Analysis

🔒

Software Security

🔍

Vulnerability Detection

⚙️

Static & Dynamic Analysis

Publications

Academic Service

03 Experience

Research Assistant

University of Texas at Dallas Aug 2023 – Present
  • Lead research in program analysis and security, applying machine learning to enhance automated vulnerability detection and testing for C/Java systems
  • Design large-scale experimental pipelines using fuzzing and dynamic instrumentation to benchmark four static analysis tools
  • Authored six papers (ICSE, ICST, and in submission to FSE/ESEM), introducing novel datasets and methodologies in static analysis research

Research Assistant

Model Driven Software Engineering Lab Jan 2022 – Jan 2023
  • Developed an automatic reservation system generator engine using MDSE approach
  • Authored a journal paper, introducing novel approach in automatically generating web applications

04 Projects

ArtCom – Online Art Auctions

WebSocket-based real-time bidding with Django & React

  • Developed online art auction platform using React.js and Django REST Framework, reducing renting costs by 90%
  • Implemented Dockerized architecture with WebSocket-based real-time auctioning supporting 200 concurrent participants
  • Designed PostgreSQL + Redis backend with user management, authentication, and data synchronization

LORD – Low-Code Reservation Platform

Model-driven system for generating reservation platforms

  • Pioneered a low-code reservation system generator that automated website creation, cutting deployment costs by over 90%
  • Engineered a scalable full-stack solution (Django/MySQL/React) with integrated payment processing

VR Physiotherapy Game

Unity VR game with ML motion tracking

  • Developed various scenes and Avatars in Unity with motions such as transportation, pulling, pushing and gravity logic
  • Architected a web-socket pipeline connecting MediaPipe model to a node.js server and pushed live notifications to the VR application

Call Graph Pruning Framework

Evaluation pipeline for reducing false positive edges of static call graphs

  • Built static call graphs (DOOP, OPAL, WALA) and instrumented Java bytecodes (ASM) to capture execution traces at runtime
  • Implemented neural networks with PyTorch and TensorFlow, automating hyperparameter tuning and fold splitting pipelines
  • Fine-tuned CodeBERT/CodeT5 models, refining attention mechanisms to improve call graph edge classification accuracy by 30%
  • Re-evaluated pruning methods, fixing dataset flaws; showed results inflated (+50% vs. actual +5%)

Explainable AI for Static Analysis

Evaluate and improve the application of ML in reducing false alarms of static analysis tools

  • Constructed the first explainable dataset for static analysis ML (1,132 Java samples, 67K C samples) evaluating 7 code LLMs
  • Improved model interpretability with Explanation-Guided Learning, boosting relevant line focus from 35% to 65%
  • Enhanced model classification accuracy with Input×Gradient xAI method by 20%

05 Get In Touch

I'm always open to discussing research opportunities, collaborations, or just having a conversation about software engineering and program analysis.

📍

Location

Dallas, Texas, USA

📧