equifax

Centelon helps one of the largest global credit bureau to redact sensitive information from the credit card images

summary

Client is one of the largest consumer credit reporting agencies. The client collects and aggregates information for 1 billion+ customers worldwide across retail and business segments.

It operates in 20+ countries in North America, Central and South America, Europe and the Asia Pacific region. It also provides background check services to employers for employee background screening, income verification, employment verification etc. 

Challenges

As part of the employee background checks, prospective employees provided scanned credit card images for verification.

As per the regulatory requirements, PII data on these images was to be redacted and obfuscated images stored for 12 months.

To achieve the requirement, client was looking for an in-premise solution which can process scanned credit card images and can mask card number with high level of accuracy. The current challenges faced in their legacy process were

  • Client had 250,000+ historical images, and masking these manually will take huge effort and are prone to human error
  • Credit cards are varied in design and typography. Hence static rules-based solutions can have high error rates
  • The proposed solution shall integrate with their website and shall mask credit card digits from the images in real-time

Solution

Based on the high-level requirements, Centelon proposed its flagship solution Metiz PII Manager for masking credit card images.

At the heart of the proposed solution, there is an AI based engine that helps clients to identify the personally identifiable information (PII) data regions and masks these regions for enhanced security. The product works on the underlying Centelon principle of Sense – Predict – Act – Learn

Predict –       Predict masking region for the credit card image

Act –               Archive masked images from the corresponding channel

Learn –          Learn from historical masked images to create / update ML models

Sense –         Sense images from multiple source viz. real-time, file, database

  • Centelon deployed PII Manager to mask all historical images in batch mode
  • Integration with client’s website where-in images captured via website are passed to the solution using REST APIs for real-time masking
  • Deployed solution to process all major types of image files viz. JPEG, PNG, BMP, TIFF, and PDF with capability to process all the major cards types viz. VISA, Mastercard, AMEX, Diners etc.
  • System has capability to mask credit card numbers, dates, and CVV numbers
  • Masked images were stored in RDMS for 12 months for regulatory audit

Benefits

  • Regulatory Compliance: Helped client achieve true PCI-DSS compliance
  • High Accuracy: Trained card number region prediction model with 99% accuracy. This highly reduced need for manual intervention
  • Quick ROI: Masked 250,000 historical card images within a week of implementation
  • Quick implementation: Centelon just took 2 months to implement PII manager for the client
  • Faster Turnaround: PII Manager masks images from website in real-time with SLA of less than 1 sec