Ali Mhrez, PhD
PhD Researcher in Data Science & NLP
I'm an NLP researcher and developer with a strong background in Deep Learning and machine learning. My work focuses on building robust models for tasks like stance detection, document summarization, and natural language inference. I am dedicated to exploring state-of-the-art techniques and contributing to the advancement of data science through research and practical application.
Professional Journey
Follow my professional journey, from foundational roles to my current research in NLP. The timeline below provides a comprehensive overview of my responsibilities and achievements in each position.
Technical Expertise
This section provides an overview of my technical capabilities, detailing the specific languages, frameworks, and foundational concepts I work with.
Languages & Frameworks
- PyTorch, Scikit-Learn, Pandas, Numpy, Matplotlib etc.
- Python, Java, PHP, JavaScript, SQL
- Laravel, jQuery, Blade, Bootstrap
- MySQL
- HTML, CSS
- Version Control (Git)
Foundational Knowledge
- Machine Learning / Deep Learning
- Natural Language Processing (NLP)
- Linear Algebra / Probability & Statistics
- Object-Oriented Programming (OOP)
- Full-Stack Development
- Software Development Lifecycle (SDLC)
Featured Projects
Here are some of the key projects I've worked on. They demonstrate my ability to lead development lifecycles and conduct in-depth research. The details for each project are shown below.
Nuanced Stance Detection Models
Using Multilingual BERT (mBERT) features as input to CNN and LSTM architectures. We explore two approaches to feature extraction: Using the last layer,and concatenating features from last four layers.
Stance Detection Using Multilingual Language Models
Investigated the potential of multilingual language models, including mbert, xlm-roberta, distilmbert, and mdeberta for enhanced stance detection capabilities.
Stance Detection Using Arabic Language Models
Explored Arabic language models, including arbert, marbert, gigabert, and albert to address specific language challenges.
Stance Detection Using Deep Learning Techniques
Implemented deep learning models, leveraging transfer learning, pre-trained embeddings, MLPs, CNNs, and LSTMs.
Insurance Record Management System
Led the full development lifecycle of a critical system for the Central Bank of Syria, from requirement gathering to deployment.
This project involved creating a comprehensive system to manage insurance records efficiently and securely. My responsibilities included:
- Gathering and analyzing user requirements from various departments.
- Designing the system architecture, database schema, and user interface.
- Leading the development and coding phases of the project.
- Overseeing the deployment, testing, and subsequent maintenance of the system.
Publications
Here are my publications.
Mhrez, A., Ramadan, W., and Saleh, N. (2024). Evaluating the Performance of Arabic Language Models on the Task of Stance Detection Toward Fake News. Journal of Homs University, 46(5), 83-106. Link
Mhrez, A., Ramadan, W., and Saleh, N. (2024). Evaluating the Performance of Deep Learning Models on the Task of Stance Detection Toward Fake News. Journal of Homs University, 46(1), 69-94. Link
Academic Background
My academic path has provided me with a strong theoretical foundation in computer science, culminating in a deep specialization in Data Science and Natural Language Processing at the PhD level.