Worked on enhancing observability for the DSA Indexer within the PaddleFormers repository, focusing on improving monitoring and reliability. Developed and integrated advanced logging for loss metrics, implemented configuration validation, and introduced robust exception handling to minimize runtime disruptions. Leveraged Python for both data logging and machine learning workflows, ensuring that loss metric logs are more actionable and less noisy. These changes enable faster troubleshooting and more effective root-cause analysis for the DSA Indexer. The work demonstrated a methodical approach to observability tooling, emphasizing maintainability and operational readiness without introducing unnecessary complexity or overhead to the existing codebase.