PaySentinel

• 

  splash screen

• 

  Landing page

• 

  real time fraud detection

• 

  how it works

Inspiration

We built PaySentinel because small merchants in India often face UPI fraud without getting help in time. Most fraud tools are designed for banks or large businesses, and many of them are hard to understand, English-only, or too slow. We wanted to create something that feels simple, fast, and useful for everyday shop owners.

What it does

PaySentinel checks UPI transactions in real time and flags suspicious activity before a merchant loses money. It gives clear explanations for why a payment looks risky and can also alert the user through voice messages in regional languages like Kannada. The goal is to make fraud detection easier to understand and quicker to act on.

How we built it

We built the backend in Python with Flask and used machine learning models to detect unusual transaction patterns. We added feature engineering to turn raw transaction data into useful signals, then used an ensemble approach to improve detection. For the user experience, we created a dashboard and added voice alert support so merchants can get warnings in a more natural way.

Challenges we ran into

One of the biggest challenges was making the system fast enough for real-time use while still keeping the detection accurate. Another challenge was making the alerts easy to understand for non-technical users. We also had to balance explainability, performance, and usability so the project would feel practical in a real merchant setting.

Accomplishments that we're proud of

We are proud that PaySentinel can detect suspicious transactions quickly, explain the reason in plain language, and deliver alerts in a regional language. We also built it in a way that is modular, testable, and ready for deployment. It feels like a complete product rather than just a model demo.

What we learned

We learned how important it is to design AI systems around real people, not just accuracy metrics. We also learned how much better a project becomes when it is explainable, accessible, and built with the end user in mind. This project helped us improve in backend development, machine learning, and product thinking.

What's next for PaySentinel

Next, we want to support more Indian languages, improve the fraud detection pipeline with more real-world data, and add stronger integrations for merchants and payment platforms. We also want to make the dashboard more polished and add more tools for fraud reports and prevention insights.

Built With

• css
• docker
• flask
• gtts
• html
• javascript
• kafka
• numpy
• pandas
• python
• pyttsx3
• scikit-learn
• shap