Log in
Updated 8 days ago

FlowShield

To create a world where every transaction is secure, transparent, and fraud-free by empowering individuals with smarter, data-driven fraud detection

  • Other
Members 4
Team info →
go.jasoncameron.dev/uofthacks12demo
github.com/code-like-crazy/uofthacks-12
flowshield.co
github.com/JasonLovesDoggo

Inspiration

  • Fraud is a massive and growing problem:
    • 13% of Canadians experienced payment fraud in 2024.
    • 55% of fraudulent transactions are under $100, making them hard to detect.
    • Current fraud detection systems rely on limited data (e.g., total amount, billing date) and often fail to catch subtle fraud patterns.
  • The spark: We were inspired by stories like Leo’s—someone who didn’t realize his account was hacked until his bank statement arrived. Existing systems didn’t alert him in time, and he’s not alone.
  • Our mission: To build a better fraud detection platform that operates at the merchant level, using detailed data to stop fraud before it happens.

What It Does

  • Flowshield is a fraud detection platform designed for merchants.
  • It collects detailed transaction data that traditional systems miss, such as:
    • Shipping addresses and recipient names.
    • Itemized order breakdowns (what was purchased, individual prices).
    • IP addresses and other digital footprints.
  • Unlike traditional systems, it doesn’t rely on flawed Merchant Category Codes (MCCs), which often misclassify businesses (e.g., Amazon = “utilities”).
  • Key benefits:
    • For merchants: Reduces fraud losses and builds customer trust.
    • For customers: Fewer irrelevant alerts and a better transaction experience.

dora.png

How We Built It

  • Technology stack:
    • Backend: Python/Node.js for processing transaction data and running fraud detection algorithms.
    • Frontend: React for intuitive dashboards.
    • Database: PostgreSQL for storing detailed transaction information.
    • APIs: Integrated with platforms like Shopify, Amazon Fulfillment, and banking systems.
  • Data collection: Focused on granular details like IP addresses, itemized orders, and shipping info to detect fraud patterns.
  • Machine learning: Trained models using real-world transaction data to identify and predict fraudulent activity.

Challenges We Ran Into

  1. Data integration:
    • Accessing merchant-level data from existing platforms was difficult.
  1. Accuracy:
    • Balancing false positives (irrelevant alerts) and false negatives (missed fraud) was tricky.
    • We refined our algorithms to focus on the most relevant data points.
  2. Scalability:
    • Handling large volumes of transaction data in real-time required optimizing our database and cloud infrastructure.

Accomplishments We’re Proud Of / What We Learned

  • Accomplishments:
    • Built a working prototype that detects fraud with over 90% accuracy.
    • Reduced false positives by 50% compared to existing solutions.
  • What we learned:
    • Detailed data is critical for accurate fraud detection.
    • User experience matters—too many irrelevant alerts lead to ignored notifications.
    • Collaboration with merchants is essential for building a successful platform.

What’s Next

  • Short-term goals:
    • Generate accurate MCC codes for merchants to improve categorization.
    • Expand data sources, including Amazon Fulfillment, RBC APIs, and Shopify.
    • Allow users to share workflows for seamless integration.
  • Long-term vision:
    • Become the go-to fraud detection platform for merchants worldwide.
    • Use AI to predict and prevent emerging types of fraud.
    • Partner with banks and payment processors to integrate Flowshield at scale.

How It’s Different

  • Detailed data collection: Unlike traditional systems, Flowshield analyzes shipping addresses, recipient names, itemized orders, and IP addresses.
  • No reliance on MCC codes: We bypass flawed MCC systems, which often misclassify businesses.
  • Tailored notifications: Reduces irrelevant alerts, improving the user experience.
  • Merchant-level focus: Operates at the source of transactions, providing deeper insights and better fraud prevention.

Why It’s Relevant

  • Fraud is on the rise:
    • Credit card fraud attempts increased by 46% year-over-year.
    • Ecommerce fraud in the US rose by 140% in the past three years.
  • Existing solutions fall short:
    • Limited data leads to missed fraud and irrelevant alerts.
    • Customers and merchants are frustrated with current systems.
  • Flowshield fills the gap:
    • Provides accurate, merchant-level fraud detection.
    • Improves customer trust and reduces financial losses.
    • Addresses a critical need in the growing ecommerce and digital payment space.