My master's thesis focused on enhancing the interpretability of BERT models, particularly in the context of Named Entity Recognition (NER) tasks. The project involved:

1. Building and Fine-Tuning BERT Models: I developed a BERT-based model tailored for the English CoNLL-2003 NER dataset, optimizing its performance for recognizing entities such as persons, organizations, and locations.
2. Token-Level Analysis: The behavior of BERT was explored at the token level to better understand how it makes predictions for individual words and their surrounding context.
3. Explainable AI Techniques: Leveraging state-of-the-art methods like LIME (Local Interpretable Model-Agnostic Explanations) and SHAP (SHapley Additive exPlanations), I visualized and analyzed the decision-making processes of the BERT model. This allowed for a clearer understanding of why certain predictions were made, contributing to trust and transparency in AI systems.
4. Evaluation and Visualization: A custom framework was implemented to evaluate and visually represent the explanations generated, facilitating actionable insights for improving model performance and usability.

This thesis not only advanced my technical expertise in deep learning and XAI but also contributed to bridging the gap between complex AI models and human interpretability.

I developed and designed a fully digital web application during my internship with GPRO-CONSULTING in Tunisia. This project aimed to create an interactive platform to connect service seekers with professionals in the automotive sector.

Key Tasks and Achievements:

1. Platform design and development:
   • Built a responsive and user-friendly interface using ReactJS , HTML , CSS , and Bootstrap , ensuring a seamless user experience.
   • Designed the backend with NodeJS , ExpressJS , and a MongoDB database for efficient data management and real-time interactions.
2. Workflow and Methodology:
   • Adopted the Scrum methodology for agile project management, enabling iterative development and frequent client feedback.
   • Created detailed UML diagrams to visualize and document system architecture and workflows.
3. Collaboration and Version Control:
   • Collaborated effectively with team members using GitHub for version control, ensuring smooth code integration and project tracking.

This project allowed me to gain practical experience in web development, problem-solving, and working with modern technologies. It also enhanced my understanding of designing scalable, interactive platforms tailored to specific industry needs.