Worked on stabilizing the CI/CD infrastructure for the pytorch/ci-infra repository by addressing a critical issue with runner availability. Using Infrastructure as Code principles and HCL, reverted a configuration change that had set the minimum available runners to zero, restoring the baseline to two and ensuring continuous integration pipelines remained reliable. This approach maintained a consistent pool of runners, reducing pipeline delays and preventing capacity gaps. The work emphasized clean change control and traceability by implementing a straightforward revert, minimizing operational risk. Focused on CI/CD reliability, the contribution improved throughput and maintained the integrity of automated build and test processes.