The Neo4j Parallel Spark Loader Python package is a solution to common issues that occur during parallel ingestion into a graph database, specifically Neo4j. It eliminates deadlocks and provides enhanced performance at high data volumes by identifying groups of relationships where no nodes are endpoints of relationships in more than one group, which can then be loaded in parallel without the risk of deadlocking. The package requires some basic knowledge of how Spark works as well as an understanding of the underlying graph structure contained in the relationships of interest. It supports three scenarios: predefined components, bipartite data, and monopartite data, each with its own module and function to create batches and groups. The package can improve load time performance around 250k row size for predefined components, while bipartite and monopartite begin to show improvement around 1.5mil rows. Additional improvements include using CREATE instead of MERGE, passing group number to ingest function, and custom property-based groups to reduce preprocessing time. The Neo4j Parallel Spark Loader Python package is useful for those looking to improve large-scale ingestion jobs by alleviating headaches caused by loading relationships in parallel and providing tools for grouping and batching data before ingestion.