Forough Mehralian
Download my resume!
Adventurous, Focused, Driven, Always up for a new challenge!
News
Mar 2024
About me
I am a PhD candidate in Software Engineering at UC Irvine, under the supervision of Dr. Sam Malek. My research interest is in automated software testing and analysis using machine learning techniques as well as data analysis. My dissertation focus is on advancing accessibility testing and repair in mobile apps. In my heart of hearts, I am passionate about automation, driven to learn from observations, and dedicated to promoting inclusion.
** I believe in giving back to the community. If you're looking for advice in areas where I have experience, feel free to schedule a meeting with me via Calendly. **
Work Experience
Machine Learning Research Intern
Mar 2024 - Aug 2024
- Designed and implemented a novel technique to assist developers in fixing accessibility issues in iOS apps.
- Designed a multi-task LLM-based agent to localize the issues in the source code and fix them automatically.
- Submitted the results of this study to ICSE 2025. View on arXiv.
Graduate Research Assistant
Jun 2019 - Sep 2024
- Designed and implemented a suite of novel automated accessibility testing tools for Android apps, utilizing machine learning techniques, as well as static and dynamic analysis methods.
- Conducted extensive research into the use of image captioning for labeling unlabeled icons in Android apps, and identified a data-driven bias in the existing approach. Developed a novel deep learning solution, which significantly improved the quality of the generated labels.
- Designed and implemented a deep learning-based technique using PyTorch for building an oracle for energy tests.
Software Engineer
Jun 2021 - Sep 2023
- Enabled web-based access to DICOM data by implementing the DICOMWEB standard. Developed Store, Retrieve, and Query APIs as a new service using REST in C++ with automated test scripts in PHP.
- Engineered a cost-efficient archive and backup mechanism in C++ for DICOM data, resulting in a remarkable 72.5% reduction in storage expenses per gigabyte, leveraging Amazon S3's storage classes.
- Enhanced defect localization by spearheading the development of the backend for the Performance Monitoring dashboard. Provided insights into critical system metrics, including hardware measures, database statistics, and DICOM-specific metrics.
Software Engineering Intern
Jun 2018 - Aug 2018
- Developed a micro-service using Django for Bazaar, Iran's largest app store with over 40 million users, to display ads on their search page. Utilized Docker and Kubernetes for containerization and orchestration of the micro-service, respectively.
Publications
Published in ICSE 2025, 47th International Conference on Software Engineering
Authors: Forough Mehralian, Ziayo He, and Sam Malek
Published in ISSTA 2024, 33rd International Symposium on Software Testing and Analysis, September 2024
Authors: Mahan Tafreshipour, Anmol Vilas Deshpande, Forough Mehralian, Iftekhar Ahmed, and Sam Malek
Published in ASE 2022, 37th International Conference on Automated Software Engineering, October 2022
Authors: Forough Mehralian*, Navid Salehnamadi*, Syed Fatiul Huq, and Sam Malek
Published in ASE 2022, 37th International Conference on Automated Software Engineering, October 2022
Authors: Navid Salehnamadi*, Forough Mehralian*, and Sam Malek
Published in ESEC/FSE 2021, 29th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, August 2021
Authors: Forough Mehralian, Navid Salehnamadi, and Sam Malek
Published in ESEC/FSE 2020, 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering, August 2020
Authors: Reyhaneh Jabbarvand, Forough Mehralian, and Sam Malek
Projects
My projects on Github.
An LSTM model with attention mechanism to automatically generate energy test oracles for mobile apps.
A context-aware label generation model with encoder-decoder architecture to generate labels for unlabeld icons.
An accessibility service that automatically explores Android apps with and without TalkBack to find overly accessible elements.