Worked on enhancing TPU job tracking and resource allocation within the apple/axlearn repository by implementing a TPU Job Labeling and Tracking Enhancement. This involved adding project-id and num-replicas labels to the TPU job configuration, which improved observability and enabled more precise resource planning for TPU workloads. Leveraged GCP and Python to instrument labeling-based tracking, laying the foundation for future cost allocation, auditing, and SLA reliability. Focused on cloud computing best practices and incorporated unit testing to ensure robust integration. The work addressed the need for better traceability and monitoring of distributed jobs, contributing to more efficient cloud resource management.