Hadjer Khaldi

Led development and deployment of multiple AI-powered systems to automate data workflows and democratize data access: – NL2SQL Query System (GeoMarketing app, CRM) – Automated Column-to-Schema Mapping (Law Enforcement Data) – Technical approach: Rigorous model evaluation, strategic prompt engineering over fine-tuning, automated pipeline development, and production-focused deployment strategies.

Improved production relation extraction system by developing and evaluating ML-based approaches to replace rule-based pattern matching methods. Key Contributions: – Analyzed existing pattern-matching relation extraction system and identified accuracy improvement opportunities – Designed and implemented neural network models (RNNs, CNNs) for relation extraction using TensorFlow – Built custom NLP pipeline leveraging spaCy and NLTK for text preprocessing and feature engineering – Constructed training datasets and evaluated model performance using scikit-learn, achieving improved accuracy over baseline pattern-matching approach.

Technology used: Angular, Ionic, Cordova plugins, Neo4J, ElasticSearch, Grapheaware, Kibana Trustiser is a multi-sided platform; it primarily manages a community of members that is arranged hierarchically. The hierarchies are built using social and big data techniques and are aimed at enabling the emergence of trusted digital advisors; they are trust-based, topic-related, and experience or expertise-linked. Specialization: Trust and reputation management, digital advisors, advisors recommandation.

Technology used: Python, Django rest API, Postgesql, Angular Conception and development of a social media information monitoring system in order to provide the decision makers, in the right time, with strategic information available in the web and put them into a structured format.