Worked on the nebius/soperator repository to implement dynamic PersistentVolumeClaim provisioning within the slurm-cluster Helm chart, focusing on automating storage configuration for SLURM workloads. Leveraged DevOps practices and Kubernetes expertise to design a Helm template that creates PVCs based on user-defined volume source configurations, reducing the need for manual intervention. Updated default values and provided commented YAML examples to guide users in specifying storageClassName and size for sources such as jail-snapshot and mlperf-sd. This approach improved deployment reliability and scalability by enabling automated, configurable storage provisioning, streamlining the setup process for complex Kubernetes-based SLURM cluster environments.