Code optimization affects a program's efficiency and maintainability and is essential to software performance. Conventional optimization efforts that rely exclusively on compilers do not address problems that arise from the software's underlying structure or recurring code patterns, which typically emerge during extensive software production. The authors present an innovative hybrid framework that incorporates automated code optimization and redundancy elimination, facilitating code quality improvement through static code analysis and clone detection. The framework finds and identifies code that is repeated or has a similar structure by looking for both syntactic and semantic similarities. Then, it uses transformation-based optimization to cut down on the extra time it takes to run. To see if our approach is suitable for Java-based applications, we have made a practical study. This study was based on some benchmark open-source software taken from the real world and on some software repositories. Our Java program was able to surpass the optimized native compiled code (less execution time by 15-25% and memory utilization by 15% mainly). Such improvements give the hybrid approach not only a clear advantage to the program’s functionality but also that the code is still easily readable and maintainable. This really opens the way for the deployment of machine learning in the field of automation software optimization.