Worked on the red-hat-data-services/kueue repository to enhance the reliability and accuracy of resource allocation within the Topology Assignment Service. Focused on backend development and system design, the main contribution involved fixing a critical bug in the assumedUsage calculation, ensuring that resource accounting now reflects actual pod usage rather than aggregate requests. This adjustment prevents misclassification of workloads and supports more precise scheduling and capacity planning. The work included implementing automated tests in Go to validate the fix and guard against regressions, demonstrating skills in debugging complex scheduling logic, resource accounting algorithms, and maintaining robust CI workflows for Kubernetes-based systems.