Eddie Bkheet enhanced performance observability in the apache/spark repository by developing a new metric that isolates actual Python execution time within PySpark UDFs, separating it from worker initialization overhead. This metric, pythonProcessingTime, was integrated into the Spark UI to provide clear visibility into Python code performance, enabling more accurate root-cause analysis and targeted optimizations for data processing workloads. Eddie’s approach involved implementing robust unit tests to validate the metric’s accuracy and ensure regression safety. Leveraging skills in Python, Spark, and performance optimization, this work addressed a key observability gap and contributed to reproducible performance KPIs for the project.