Inspiration

Online payments are growing rapidly, but fraudulent activities are becoming increasingly sophisticated. Traditional fraud detection systems often analyse transactions individually and may fail to identify coordinated fraud involving multiple interconnected accounts.

We were inspired to build a smarter payment security solution that can identify suspicious relationships and detect potential fraud networks using Artificial Intelligence and Machine Learning.

What it does

PayGraph is an AI-powered payment fraud detection system that uses Graph Machine Learning to analyse relationships between payment accounts and transactions.

Key features:

  • Graph-based detection of suspicious transaction networks.
  • Machine Learning-based anomaly detection.
  • Intelligent transaction risk scoring.
  • Explainable AI to identify reasons behind suspicious activity.
  • AI-powered investigation assistance for merchants.

How we built it

The system integrates the PayPal Developer Sandbox to simulate payment transactions and collect relevant payment events.

We use Python-based Machine Learning algorithms and graph-based learning techniques to analyse transaction relationships, identify anomalies and generate risk scores. An AI assistant helps interpret suspicious patterns and generate understandable investigation summaries.

Challenges we ran into

Some of the main challenges include representing payment transactions as interconnected graph data, identifying meaningful fraud patterns, integrating PayPal APIs and combining graph-based Machine Learning with explainable AI.

Accomplishments that we're proud of

  • Developing an innovative approach to payment fraud detection by analysing interconnected transactions rather than examining individual payments alone.
  • Combining Graph Machine Learning, anomaly detection and Explainable AI into a unified fraud intelligence solution.
  • Designing a system that can identify suspicious transaction relationships and highlight potential coordinated fraud networks.
  • Integrating AI-powered investigation assistance to make complex fraud patterns easier to understand.
  • Exploring how PayPal’s payment ecosystem and advanced AIML techniques can work together to improve digital payment security.

We are especially proud of turning a real-world financial security challenge into an AI-driven solution with the potential to make online payments safer and more intelligent.

What we learned

Through this project, we aim to gain practical experience in Graph Neural Networks, anomaly detection, explainable AI, payment API integration and intelligent financial security systems.

What's next for PayGraph-AI-powered payment security agent

We aim to extend PayGraph with adaptive fraud detection, real-time transaction monitoring and more advanced graph learning models to improve payment security.

Built With

Share this project:

Updates

Submission history