Nassira Nguadi

Industrialization of a synthetic identity document generation pipeline to create a dataset covering 15 classes (passport, national ID card, health insurance card, driving licence, etc.), with more than 1,000 samples per class. The work includes training and evaluating a detection and entity extraction model, benchmarking it against existing state-of-the-art models, and experimenting with alternative architectures using both real and synthetic datasets.

Design and development of advanced information extraction and computer vision solutions, including implementation of a model for extracting 10 types of named entities and their relationships, development of a computer vision system capable of detecting 15 distinct types of links, as well as fine-tuning a model specialized in structured text extraction from resumes, with automatic generation of results in JSON format.

Internship at CapitalTech-Rabat, Morroco 07/2024–10/2024 Designed solutions to assess CV quality by evaluating the completeness of sections, analyzing the relevance of content, and identifying spelling and grammar errors to ensure high standards in CV presentation and accuracy. Implemented intelligent search within CVs by using Named Entity Recognition (NER) to identify key entities, extracting relation ships between entities, and enhancing the accuracy and relevance of CV data retrieval.

Designed and developed a library management application, enabling efficient cataloging and tracking of books, streamlining borrowing and return processes, and providing users with a user-friendly interface for easy access to library resources.