AI-Powered Testing Frameworks for Complex Software Systems

Abstract

This paper presents an advanced AI-powered testing framework aimed at optimizing the software quality assurance (QA) lifecycle for complex and modular software systems. Building upon the foundational work of Kothamali and Banik [1], who introduced a machine learning-based approach for predictive defect tracking and proactive risk management, our framework extends their methodology by incorporating real-time code coverage analysis, intelligent test case prioritization, and adaptive feedback mechanisms. These enhancements allow the testing process to evolve dynamically based on live system behavior and test outcomes. The proposed framework was implemented and evaluated across two real-world environments: a microservices-based banking platform and a cloud-native Human Resource Management (HRM) system. In both cases, the system demonstrated significant improvements in fault localization accuracy, reduction of redundant test executions, and early detection of high-risk modules. Moreover, the integration of continuous learning principles enabled the test suite to adapt over time, leading to improved test efficiency and greater code coverage with fewer resources. This study confirms the extensibility, robustness, and ongoing relevance of the original framework by Kothamali and Banik [1], especially when applied to scalable, service-oriented software architectures. By incorporating AI-driven decision-making and feedback loops, the enhanced framework paves the way for more intelligent, self-improving QA practices in today’s fast-paced development ecosystems.