![case study](https://cdn.quantiphi.com/2024/03/ffc8f7fd-payments-transformation-modulr-poc-cs-banner-image.png)
Business Impacts
Real-Time Fraud Flagging and Benchmarking
Multi-Tenancy for Rule Suites, UI-Based Case Management, Case Analytics, and Dashboarding
Explainable AI for Fraud Detection
Customer Key Facts
- Location : Europe
- Industry : Financial Services
Problem Context
The client is a leading digital payment as a service provider offering businesses a faster, easier, and more reliable way to move money. The client sought to develop a real-time fraud detection system with low false positives for their Point of Sale transactions.
Challenges
- Need of real-time Fraud Detection at Point of Sale
- The model needed to be explainable
Technologies Used
![Amazon API Gateway](https://cdn.quantiphi.com/2024/03/Amazon-API-Gateway_light-bg@4x-1.png)
Amazon API Gateway
![AWS Lambda](https://cdn.quantiphi.com/2024/02/83e09c49-aws-lambda.png)
AWS Lambda
![Amazon Aurora](https://cdn.quantiphi.com/2024/02/4cf9c3f7-amazon-aurora-1.png)
Amazon Aurora
![AWS S3](https://cdn.quantiphi.com/2024/02/9c40e5d8-amazon-s3.png)
AWS S3
![Amazon Sagemaker](https://cdn.quantiphi.com/2024/02/50f0a997-arch_amazon-sagemaker_64@5x-1.png)
Amazon Sagemaker
![Amazon Fraud Detector](https://cdn.quantiphi.com/2024/03/Amazon-Fraud-Detector@5x.png)
Amazon Fraud Detector
Set Up a Fraud Detection Platform for Transactions
Solution
Quantiphi helped them set up a fraud detection platform for transactions at Point of Sale. Salient features of our solutions included:
-Client onboarding/ Risk Scoring: Conduct historic screening of a new client across millions of data items to feed into their risk score
-Real-time Monitoring: Screen millions of clients transactions across billions of data items in real-time to identify sanctions, politically exposed accounts, and adverse media.
-Threat Investigations: Connect the dots via an all-source investigations tool for exploring screening results.
-AI-driven Anti Money Laundering: Verify applicant identity, maintain verification records, and flag in real-time when a person appears on any suspected criminal list.
-Anomaly Detection: Deploy self-learning ML models trained to recognize well-known fraud patterns as well as to adapt to new, unknown fraud techniques.
Result
- The new platform had real-time fraud flagging and benchmarking capabilities, API Endpoints for fetching risk scores, and explainability for flagged transactions. In addition, it has a rule engine and API Endpoints integrated with the current pipeline