Bevan Smith

Dr. Bevan Smith is a Senior Lecturer at the School of Mechanical, Industrial & Aeronautical Engineering at the University of the Witwatersrand (Wits), where he also completed his BIng(Mech), MSc(Eng), and PhD. His research bridges machine learning and engineering systems, with a focus on optimization through simulation and AI.

His research interests include:

• Machine Learning and Reinforcement Learning
• Causal Machine Learning and Causal Discovery
• Graph Neural Networks
• Explainable AI (XAI)
• Digital Twins
• Application of AI in Education and Industrial Systems

His recent publications demonstrate a strong trajectory in applying causal inference and machine learning to both industrial optimization (e.g., inventory management, air quality monitoring) and educational challenges (e.g., at-risk student prediction, online video interventions). He utilizes simulation-based methods and counterfactual reasoning to evaluate treatment effects and improve decision-making. His work spans journals in industrial engineering, artificial intelligence in education, and computational applications.

Notable scientific contributions include:

• Developing AI-driven solutions for low-cost air quality monitoring (AI_r)
• Evaluating X-Learner performance under confounding and non-linearity
• Applying Pearl’s counterfactual framework to assess explanation validity
• Designing personalized interventions for engineering students using model-agnostic explanations

Dr. Smith has contributed to curriculum development through the creation of online educational videos and has explored structural design in mechanical engineering, notably in carbon fibre swingarm development. While no formal advising or grant information is available, his work reflects an interdisciplinary approach combining mechanical engineering principles with cutting-edge AI methodologies. He is actively publishing and remains engaged in both technical and educational research domains.