The goal of this thesis is to design and implement scavenger framework for Kubernetes. The framework will use idle Kubernetes cluster resources to run additional workloads and will maintain set utilization level of the entire cluster. That is, it will run additional workloads when the overall cluster utilisation is below a specified threshold, and remove workloads if the utilization is above a specified threshold. Student will explore Kubernetes development patterns and components (Operator pattern, controllers, Webhooks) and implement a new custom resource representing the scavenger workload and a dedicated management system. Significant attention will be paid to selecting a proper workload for removal so that computing resources are wasted minimally. Some preliminary tests and comparison with existing solutions such as Boinc shall be included in the thesis. Expected output is a prototype implementation and a thesis text to serve as documentation.