The objective of the proposal is to design deep learning-based graph embedding and clustering techniques to detect the circular trade in the stock market and GST. The aim is to design algorithms to detect fraudulent GST returns and Money Laundering detection by detecting layering, identification of mule accounts.