Worked on the NVIDIA/doca-platform repository to enhance DPU management and reliability in a Kubernetes environment. Delivered a feature that validates DPU readiness before allowing node reboot updates, ensuring updates proceed only when all DPUs report a Ready state and reducing the risk of hardware instability. Implemented comprehensive tests for readiness scenarios and updated webhook validation logic to align with these checks. Additionally, addressed a bug in the DPUNode deletion workflow by ensuring associated DPU and DPUDevice resources are properly cleaned up, preventing resource leaks. All work was completed using Go, with a focus on backend development and cloud infrastructure stability.