An AI-based service model for optimizing career development paths of STEM students

Abstract

In response to the problems of insufficient personalization and low prediction accuracy in career development path planning for science and engineering students, this paper constructs an artificial intelligence (AI) algorithm based career development path optimization model. At the input data level, the model integrates multidimensional historical data of science and engineering students, including human capital indicators such as academic performance and mastery of professional skills, as well as social capital indicators such as social activity participation, internship experience, and mentor evaluation. At the logical level of path generation, the model first uses the Random Forest (RF) algorithm to rank the importance of high-dimensional features and select the optimal features, in order to screen out the key variables with the strongest predictive ability for students' employment destinations; On this basis, a hierarchical prediction framework is constructed using a deep belief network (DBN), which explores the complex nonlinear mapping relationship between features and career development outcomes through layer by layer unsupervised pre training and supervised fine-tuning. The system outputs personalized career development path recommendation plans, including target industry adaptability, expected development prospects, and phased skill improvement suggestions. The experimental results show that among 200 prediction task samples, the prediction accuracy of this model reaches 84.2%, and the recall rate and user satisfaction are significantly better than traditional support vector machine models, verifying the effectiveness and practicality of this model in predicting the career development of science and engineering students.