Worked on the opendatahub-io/opendatahub-tests repository to enhance MLServer by implementing Persistent Volume Claim (PVC) storage support, enabling models to be stored and accessed from PVCs rather than relying on S3. Used Python and Kubernetes to migrate model storage, introduce fuzzy validation for model integrity, and remove outdated snapshot files, thereby improving test reliability and data consistency. Standardized deployment mode naming to align with KServe conventions and resolved fixture naming conflicts. Added comprehensive end-to-end test coverage for PVC-based model storage, streamlining CI processes and reducing cloud dependency while focusing on robust machine learning testing and validation workflows.