Worked on the nebius/soperator repository to enhance reliability by addressing two high-severity bugs in the operator lifecycle. Focused on backend development using Go, Kubernetes, and Python, the work involved improving worker initialization by handling unregistered nodes during the scontrol update process, which prevented system crashes and increased robustness. Additionally, container rendering stability was strengthened by guarding the ProcMount field against empty strings in Kubernetes contexts, thereby eliminating potential nil pointer errors. These targeted fixes reduced runtime errors and on-call incidents, contributing to smoother upgrades and improved cluster stability while demonstrating effective Go-based operator development and Kubernetes integration.