Abstract
The rapid acceleration of technological innovation poses a significant challenge to higher education institutions, particularly within engineering disciplines where traditional curricular revision cycles often lag behind industry requirements. This study introduces a predictive analytics framework utilizing natural language processing (NLP) and deep learning architectures to forecast longitudinal skill demands and optimize engineering curricula. Analyzing an empirical dataset of over 2.4 million engineering job postings collected across five years alongside historical syllabus repositories, we deployed a hybrid Bidirectional Encoder Representations from Transformers (BERT) model coupled with a Long Short-Term Memory (LSTM) network to track competency trends and project demand trajectories. Our findings indicate an accelerating emergence of hybrid competencies—notably integrating artificial intelligence operations, cyber-physical systems security, and sustainable systems engineering—into foundational engineering profiles. When applied to an ABET-aligned multidisciplinary engineering curriculum, the model successfully identified structural competency deficits and generated optimized course modification pathways that reduced the projected curricular misalignment score by 34.6%. By bridging the gap between real-time labor market telemetry and educational program design, this framework offers academic stakeholders a data-driven mechanism to proactively modernize engineering education.
