In today’s digital age, industries demand increasingly higher software quality. As a critical component in ensuring this quality, software testing plays an indispensable role. However, the current test case design still heavily depends on testers’ personal experience and theoretical understanding, leading to uncertainty in testing outcomes and affecting reliability. This paper thoroughly explores software quality assurance (SQA) strategies under the DevOps framework and proposes an innovative machine learning (ML)-based automated defect detection method. The proposed approach utilizes supervised learning models trained on historical defect datasets to automatically identify, classify, and predict potential defects during the testing phase. Experiments conducted on real-world enterprise systems show that the method improves test coverage by 12%, reduces missed defect rates by 18%, and shortens test cycles by 25%, compared with traditional testing methods. This work offers a practical and intelligent solution to enhance software quality assurance within DevOps environments.