Kubernetes v1.35 introduces significant improvements to scheduling, particularly with the new Workload API aimed at optimizing the management of large workloads. Traditional scheduling often treats each Pod independently, which can lead to inefficiencies, especially in complex scenarios like machine learning jobs where Pods need to be co-located. The new features make workload scheduling a priority for the kube-scheduler, recognizing its importance in managing AI-related tasks and other large-scale applications. The Workload API allows users to describe the scheduling requirements for groups of Pods. A notable feature is gang scheduling, which ensures that all Pods in a group are scheduled together or not at all. This prevents resource wastage from partially scheduled jobs. Users can define configurations, such as a minimum count of Pods that must be scheduled together, enhancing the efficiency of resource utilization. The implementation includes a system where Pods wait at a Permit gate until the conditions for their group are met, preventing them from being scheduled until all requirements align. Another addition is opportunistic batching, which improves the scheduling speed for identical Pods without requiring user intervention. The scheduler can reuse feasibility calculations for Pods with the same specifications, significantly reducing latency. However, this feature only works when all scheduling parameters are identical. Users may need to check their kube-scheduler configurations to ensure batching is enabled. The article outlines future goals for workload-aware scheduling improvements, including introducing a dedicated scheduling phase, enhancing multi-node scheduling, and better integration with autoscaling. These ongoing developments aim to streamline workload management in Kubernetes, making it more efficient and responsive to user needs.