Artificial intelligence has revolutionized quality assurance in industries by detecting defects and anomalies, reducing costs, and protecting reputations. Traditionally, a model-centric approach focused on improving AI performance through hyperparameter tuning. However, a shift towards data-centric AI emphasizes enhancing system performance by improving data quality and relevance. This paper explores the fundamentals of both approaches, highlighting their application in quality assurance. It reviews deep learning methods, implementation challenges, and proposes a data-centric pipeline for robust quality control. By refining data continuously, this approach can minimize errors and improve manufacturing processes, offering valuable insights into advancing quality assurance with data-centric AI.