Alumni Multidimensional Job Analytics with AI agents

Abstract

Understanding and tracking alumni career trajectories is increasingly important for academic in- stitutions, students, and employers, yet integrated, actionable data remains scarce. This thesis presents the development of Alumni-FEUP, a platform designed to bridge this gap by analyzing the career outcomes of alumni from Informatics Engineering, Software Engineering, and other degrees managed by the Informatics Engineering Department at FEUP. The system integrates data collection, AI-driven enrichment, and interactive visualization to provide a multidimensional perspective on alumni outcomes. Key features include the use of ded- icated AI agents to classify professional roles, estimate seniority levels, and resolve ambiguous location data, leveraging large language models and retrieval-augmented pipelines for adaptabil- ity in a dynamic job market. The platform supports analysis across several dimensions, such as professional roles, seniority, geographic distribution, academic background, and temporal career progression. Validation combines both quantitative and qualitative approaches. Quantitative metrics include the proportion of alumni covered, automated classification accuracy for job roles, seniority, and location (validated by user feedback and manual review), and user engagement rates on the plat- form. Qualitative analysis draws on open-ended user feedback and System Usability Scale (SUS) responses from a live deployment. Results indicate strong engagement, meaningful insights for curriculum alignment, and validation of the enrichment pipeline's effectiveness. The project is publicly accessible, open source, and intended as a model for similar institutional initiatives.