Nada Arfaoui

Migrated a legacy semantic matching engine to Infor's standardized cloud ML deployment template, resolving dependency and packaging issues. Diagnosed a production memory-pressure failure and redesigned the model storage layer, cutting inference memory to ~500MB and scaling the engine from a 200K-record failure point to 900K records.

Gained hands-on experience in Python for data preprocessing, EDA, feature engineering, and data visualization. Applied supervised, unsupervised, and deep learning models to real-world datasets and delivered AI-driven capstone projects.

Developed a full-stack web application to automate administrative workflows, reducing manual data entry by 70%. Designed and optimized database schemas, implemented RESTful APIs, and reverse-engineered legacy systems for integration.

Developed an AI-powered chatbot using Azure AI Foundry with fine-tuned models to deliver personalized and context-aware customer support.

Developed and deployed a machine learning model to predict depression risk using data from the XAPI. Integrated the predictive model into a Telegram chatbot, focusing on combining NLP and predictive analytics for early risk identification.

Completed hands-on training in data analysis and worked with modern tools to analyze real-world project data.

Pursuing a Software Engineering Degree at South Mediterranean University, the first Tunisian ABET-accredited university, as a full scholarship student and Thomas Jefferson Scholarship finalist.

Earned a Bachelor's degree in Business Intelligence from the Higher School of Digital Economy.