In your role as a system architect for analyzing seismic data, you have designed an ETL (Extract, Transform, Load) process utilizing a series of MapReduce jobs within an Apache Hadoop cluster. Currently, your ETL process takes several days to complete due to the computationally intensive nature of certain steps. Recently, you identified that a crucial step for sensor calibration was overlooked. How should you modify your ETL process to ensure that sensor calibration is systematically included in future operations?