EMNA BEN JAZIA

• Designed and implemented an unsupervised model for scoring the semantic relevance of French news articles against a set of predefined contexts and subcontexts. • Conducted extensive research and experimentation with multiple NLP techniques (Cosine Similarity, Soft Cosine, Word Mover’s Distance, clustering methods, SBERT/MiniLM). • Created, refined, and maintained a comprehensive dictionary of contexts, ensuring maximum coverage while keeping subtopics semantically distinct. • Gained hands-on experience in word embeddings, semantic similarity, and NLP pipelines, while honing coding skills and applying rigorous problem-solving under strict constraints (lightweight models, no training dataset, no deep learning).

As a full-stack web development intern at SeekMake, I contributed to both quality assurance and feature development. I began by conducting detailed QA testing across the platform, delivering daily reports that identified bugs, UX issues, and potential improvements. Later, I developed a real-time statistics dashboard for the admin portal, building backend services with NestJS and MongoDB to aggregate platform usage data—including quotes, orders, drafts, service distribution, and 3D printing trends—and visualizing it with Angular and ngx-echarts.

I participated in API orchestration as well as testing and logging implementation