With the growing use of embedded systems in various industries, the need for automated platforms for the development and deployment of customized Linux-based operating systems has become more important. This research was conducted with the aim of designing and implementing an integrated and reproducible infrastructure for the development, building, and testing of a Linux-based operating system using the Yocto Project. The proposed structure was implemented based on a three-layer architecture consisting of the main Yocto repositories, a custom layer (meta-custom), and a coordinating manifest layer to ensure version synchronization, scalability, and reproducibility. Three sample projects, including libhelloworld, helloworld, and the kernel module hello mod, were developed and integrated into the build process. Continuous Integration and Continuous Deployment pipelines were implemented with GitLab CI and combined with an isolated Docker environment to automate and streamline the build and testing workflows. Using a local cache server containing hashserv, downloads and sstate cache significantly reduced the build time. The functionality and stability of the system were verified through six boot test scenarios in the QEMU simulator. The results show that the proposed design not only ensures reproducibility but also can be extended to advanced applications such as continuous deployment of real-time Linux versions. Future recommendations include expanding automated tests, implementing system monitoring with Prometheus and Grafana, using distributed builds, optimizing with Docker multi-stage builds, and enabling continuous deployment of real-time Linux changes to provide a stable and scalable model for industrial and research projects in embedded systems with a rapid and reliable development cycle.