DataGuard Investigator

🛡️ What is DataGuard Investigator?

DataGuard Investigator is an AI-powered fraud detection and investigation tool designed to help identify suspicious financial transactions and make investigations faster and easier.

💡 Inspiration

Fraud investigations can involve reviewing many different signals before deciding which transactions deserve attention. We wanted to build a simple tool that could bring those signals together and provide investigators with a clear risk assessment.

🔍 How it works

DataGuard analyzes transaction information such as:

• Transaction amount
• Account age
• Failed login attempts
• New device activity
• Transaction location
• Suspicious location activity

The system calculates a risk score from 0–100 and classifies the transaction as Low, Medium, or High Risk.

It then uses Gemini to generate an AI investigation explanation that describes why the transaction received its score, which signals are concerning, and what an investigator should check next.

🤖 Built with Gemini

Gemini is used to turn the detected risk signals into a clear, human-readable investigation explanation. This allows an investigator to understand the reasoning behind the risk assessment instead of seeing only a number.

🛠️ How we built it

The application was built with Python and Streamlit. A rule-based risk engine evaluates transaction signals, while Gemini provides the investigation explanation. The Gemini API key is securely stored using Streamlit Secrets.

📚 What we learned

We learned how to combine rule-based fraud detection with generative AI, connect a Streamlit application to the Gemini API, securely manage API credentials, and design an interface that presents investigation results clearly.

🚧 Challenges

One of the biggest challenges was integrating the Gemini API and troubleshooting model availability and Python indentation issues during development. We also tested different transaction scenarios to make sure Low, Medium, and High Risk results worked correctly.

🎯 Our goal

DataGuard Investigator is designed to assist human investigators, not replace them. Its purpose is to turn suspicious transaction data into useful investigation insights and help investigators decide what to examine next.

Built With

• au
• cybersecurity
• detection
• fraud
• gemini
• google
• machine
• python
• streamlit