Google Cloud Vertex AI and Neo4j are used together to build graph-based machine learning models. Graphs describe the relationships between data points, making them suitable for handling complex business problems. Google Cloud and Neo4j offer scalable tools for leveraging graph data. The PaySim dataset is used as an example to identify fraud in credit card transactions. Data scientists can use basic graph algorithms or generate graph embeddings to encode complex graph data into values that tabular machine learning algorithms can use. The embeddings are then combined with Vertex AI to train a supervised model. Future work includes exploring improved datasets, boosting and embedding methods, and manually engineering new graph features.