First-Time UX
Logo

Effipay: AI-Powered Payment Routing Engine

The Problem

Meet Jeff. Jeff is like most of us. He doesn't have time to wade through endless credit card tier list videos or decode reward programs as if they were complex API documentation. Like many busy individuals, Jeff faces the daunting task of selecting the best credit card from a sea of options—all while managing... well, life!

Our Inspiration

We were inspired by real-life challenges that everyday users like Jeff face. The complexity of navigating credit card rewards and payment options felt outdated in a world that thrives on simplicity and smart technology. Our vision was to leverage machine learning and AI to turn this cumbersome process into a seamless, personalized experience.

What We Built

Effipay is a comprehensive solution that streamlines payment processing and maximizes rewards, all through an intelligent, AI-driven system. Our project features:

ML-Powered Dynamic Payment Routing Engine: Determines the optimal payment methods dynamically based on spending patterns.
AI Recommendation System with MongoDB Vector Search: Delivers personalized credit card suggestions by analyzing user data.
Centralized Payment Card: Simplifies the checkout process by consolidating multiple payment methods into one streamlined interface.

User Journey

1. First-Time Signup & Onboarding (Next.js)

Account Creation: Users start by creating a new account.
Preference Setup: They are prompted to select their reward preferences.
Data Authorization: Users grant access to their credit card information and transaction history.
Initial Recommendation: An LLM processes this data and provides an initial recommendation (e.g., a Capital One Quicksilver credit card).

2. Seamless Checkout Experience (Next.js)

Mock Web Store Checkout: When shopping online, users click on the “Pay with Effipay” button.
Payment Split Recommendation: They are redirected to our site to view a tailored recommendation for splitting payments.
Transaction Completion: After authorizing the split payment, the purchase is successfully processed.

3. Interactive Insights Dashboard

Dashboard Overview: Users can access a centralized dashboard that tracks earnings towards various reward types.
Chatbot Assistance: A built-in chatbot helps users query recommendations for credit progression or expenditure optimization.

How We Built It

Frontend: Developed using Next.js to create responsive and dynamic web interactions.
Backend & Data Management: Utilized MongoDB’s vector search capabilities to power our AI recommendation system.
ML & AI Integration: Created multiple ML pipelines using a fine-tuned BERT-based embedding model for our Vector Search indexes, as well as large language models to process user data and generate personalized credit card recommendations.

Challenges We Faced

Model Tuning: Building our machine learning models to deliver accurate and meaningful recommendations. Specifically, fine-tuning our vector embedding model proved quite cumbersome.
Security & Privacy: Maintaining stringent security standards while handling sensitive user data and ensuring compliance with data protection regulations.

Conclusion

Our aim was to empower users like Jeff to effortlessly manage their payments and maximize their rewards, and we think we've done just that. We’re excited to share our journey and look forward to pushing the boundaries of fintech innovation!

Video submission: https://youtu.be/27joNbyRzZA