1Mr.Jajjara Bhargav, 2Kambhampati Bhavya, 3Unnam Jeevan Datta, 4Gowtham, 5Mujahid Shaik 03 04 The rapid adoption of Unified Payments Interface UPI systems has revolutionized digital transactions by enabling instant secure and convenient money transfers However the exponential growth of digital payments has also led to a significant rise in fraudulent activities posing serious threats to financial security Traditional fraud detection mechanisms rely on manual verification or rulebased systems which are often inefficient and incapable of identifying sophisticated fraud patterns in real time This research presents a Machine Learningbased UPI Fraud Detection System designed to identify fraudulent transactions efficiently and accurately before completion The proposed system analyzes transaction data such as transaction amount time location account behavior and device information to detect anomalies By leveraging machine learning algorithms the system learns patterns from historical transaction data and classifies transactions as either normal or fraudulent The system is implemented using Python with Flask for web application development and SQLite for database management The realtime processing capability ensures that fraudulent transactions are detected instantly minimizing financial losses The architecture of the system consists of data collection preprocessing feature extraction model training and prediction modules The system continuously monitors transaction behavior and identifies suspicious patterns using trained models Upon detection of a fraudulent transaction the system generates alerts and stores transaction details in the database for further analysis Experimental results demonstrate that the system achieves high accuracy and low error rates making it suitable for realworld applications The system was tested with multiple transaction datasets and achieved an accuracy of 94 with minimal response time The proposed approach significantly enhances transaction security and reduces the dependency on manual monitoring This research contributes to the development of intelligent fraud detection systems by integrating machine learning techniques into digital payment platforms Future work includes the incorporation of deep learning models realtime cloud deployment and advanced behavioral analytics to further improve detection accuracy and scalability Overall the system provides a robust and efficient solution for secure digital transactions Copyright (c) Journal of Science Engineering Technology and Management Science. All rights reserved 2026