HackSoda_Retailer-Fraud-Detection

Inspiration

The rise in counterfeit and fraudulent suppliers in remote locations has created challenges for eCommerce platforms, impacting both customer safety and seller reputation. HackSoda_Retailer-Fraud-Detection aims to address these issues, inspired by the need for trust in global supply chains.

What it does

HackSoda_Retailer-Fraud-Detection uses the Mistral-large-latest LLM on AWS EC2 to analyze supplier data and identify potentially fraudulent manufacturers or wholesalers. This helps sellers make informed sourcing choices and protects customers from unreliable products.

How we built it

We developed the project using Streamlit with Python for the frontend and integrated the Mistral-large-latest LLM model on an AWS EC2 instance for backend processing. This setup allows for real-time insights and decision support in fraud detection.

Challenges we ran into

Initially, procuring dataset was a major challenge since there is not any publicly available dataset. Integrating the large LLM model on AWS while ensuring scalability and quick response times was challenging. Additionally, balancing data privacy with real-time fraud analysis posed complexities.

Accomplishments that we're proud of

We successfully deployed a robust fraud detection system that can handle large-scale data and provide accurate insights to sellers. This tool empowers eCommerce platforms to secure their supply chain effectively.

What we learned

We gained experience in deploying advanced LLMs on AWS, learned more about supply chain vulnerabilities, and refined our understanding of real-time fraud detection challenges and solutions.

What's next for HackSoda_Retailer-Fraud-Detection

Future plans include expanding the model to detect fraud patterns in a wider range of industries and improving detection accuracy by integrating more data sources for comprehensive fraud analysis. In addition to this, we can scale to detect scams from the customers' end such as return frauds etc.