AI-Driven Unit Test Generation

In the context of cloud-based software development on Amazon Web Services (AWS), adopting DevOps practices through services like AWS CodePipeline is essential for ensuring Continuous Integration (CI) and Continuous Delivery (CD) while maintaining high software quality standards. However, integrating automated testing within these pipelines can be complex, particularly when aiming to generate comprehensive, high quality tests with good overall coverage. To address this challenge, this study proposes leveraging AI-driven test generation techniques within DevOps pipelines on AWS CodePipeline, followed by a manual approval phase for the generated tests. The primary objective of this research is to explore the use of Artificial Intelligence (AI) for automated test generation to optimize DevOps practices, enhance software quality, and accelerate release cycles. The study focuses on designing a pipeline within AWS CodePipeline, incorporating all key Software Development Life Cycle (SDLC) stages (source, build, test, and deploy), while identifying automated test generation as a crucial area for AI integration. The required infrastructure will be provisioned using AWS CloudFormation, an Infrastructure as Code (IaC) service. The experimentation will primarily target microservices architectures (Java, Spring Boot), which will serve as the foundational source for this thesis. The proposed approach involves integrating AI-powered tools such as OpenAI or custom solutions into the pipeline architecture to generate automated tests. These generated tests will then undergo an approval phase to assess their quality, relevance, and adequacy, ensuring a qualitative control over the generated test suites. The results of this research aim to demonstrate the potential of AI-driven test generation techniques in improving software quality, development productivity, and the release process of new components, while fully embracing DevOps best practices.