Cong worked on the DataDog/cilium repository, delivering an Azure IPAM Subnet Discovery Optimization that introduced a targeted three-phase subnet discovery process. By reworking the approach to query only subnets actually used by nodes, Cong reduced unnecessary API calls and throttling, improving both scalability and reliability for Azure onboarding. The implementation, written in Go, replaced broad VNet enumeration with focused subnet queries, resulting in lower API load and faster subnet discovery. Throughout the project, Cong demonstrated skills in API development, cloud infrastructure, and networking, while maintaining backward compatibility and clear commit practices to ensure maintainability and ease of future development.