An AI-driven, real-time career guidance framework for enhancing the employability of Sri Lankan IT undergraduates

Abstract

The changing requirements of the Sri Lankan IT industry continue to affect Sri Lankan graduates’ employment prospects due to an insufficient alignment between the skills of IT undergraduates and the industry’s needs. This paper aims to fill this gap by explicitly outlining the skill gap among IT undergraduates and industry demands, and by introducing a new AI-based framework integrating real-time actual labour market data to facilitate adaptive and personalized career guidance. It seeks to answer the following three research questions: (1) What current career guidance frameworks are most critical to address? (2) What emerging technological capabilities enable personalisation and real-time relevancy? (3) What elements ensure the framework aligns with critical and evolving IT position requirements? A systematic literature review and initial data-driven analysis were carried out using IEEE Xplore, ResearchGate and Elsevier to strengthen the methodological strength and justify the design choices of the proposed framework. Thematic analysis was applied to fifteen highly relevant studies grouped as existing frameworks, technological solutions and present concerns. Findings indicated that while AI-based systems improved personalisation, they lacked real-time labour market integration, dynamic responsiveness or empirical grounding, highlighting the originality and salience of the proposed framework’s data-based, real-time approach. Traditional systems do not rigorously analyse competencies to create tailored, precise target-configurable recommendations. The need for predictive analytics for this specific system, contextualised industry collaboration and the evaluation of system-wide competencies, as well as industry-wide competencies, was particularly noted as an area for improvement. AI-powered career guidance frameworks are necessary, and it is proposed here that they utilise Natural Language Processing (NLP) for CV analysis, machine learning algorithms such as Random Forest and SVM for algorithmic role assignment, as well as continuous monitoring of the labor market. Such frameworks would aim to provide and constantly refine tailored suggestions through responsive industry adaptation. The next phase will include prototype creation, pilot testing, and verification through user surveys to assess employability impact and productivity increase, addressing future work suggested by reviewers. This work seeks to provide an adaptable and context-sensitive model to improve decision-making in career choice, thereby enabling IT graduates to better manage the intricate and rapidly evolving pathways of their professional lives.