Credit scoring solution for automated credit decision in the cloud

FinTech providers must identify patterns and signals in a customer’s financial behavior to provide deeper, up-to-date insights into their lending assessment and credit risk. Lenders use these insights to improve decision-making and customer management capabilities. Credit scoring models and algorithms play a significant role in predicting, automating, and classifying customers according to risk profiles from credit history data.

This blog demonstrates how FinTech providers can introduce automated decision-making into their customer’s digital onboarding journeys by deploying credit scoring models on AWS.

How to manage credit profiling and risk categorization in the lending decision

With the advent of Neo banks and Open Banking, onboarding journeys are increasingly digital. Most fintech providers expect most of these journeys to be initiated and completed online. A case in point is consumers looking to apply for credit. The customer journeys typically involve credit providers performing credit profiling and risk categorization to ensure loans are offered to consumers with a low likelihood of default. Fintech providers deploy credit scoring models to predict the probability of default.

1. Historical transactional data: Historical transactional data for current borrowers are typically sourced from on-premises data sources into a data lake in the cloud. It is then ingested and transformed for downstream analytics. Additional transaction information for other credit facilities not held by the lender can be sourced from credit bureaus and other sources.
2. Create feature store: Background processes run on the transactional data to create aggregated feature variables used by the scoring algorithm.
3. Consumers apply for credit: Consumers use the FinTech provider’s front-end mobile or web channel to apply for credit.
4. Real-time scoring: Credit score is generated in real-time using the credit scoring algorithm deployed to deliver API endpoints to the consumer-facing digital channel
5. Credit decision on the edge: The channel app utilizes these insights to decide on loan eligibility on a risk-based rule engine to drive the customer journey.

How to implement and deliver the credit scoring model in AWS

1. Data Ingestion: Custom python scripts are used to replicate the on-premise historical transactional data to Amazon S3. FinTechs can source this transactional data into a highly available and scalable data lake solution on AWS.
2. ETL & Storing: Subsequently, AWS Glue is used as a serverless data integration service, to cleanse, prepare, and transform transactional data into formats supported by the AWS Redshift data warehouse.
3. On-demand scoring: The scoring models can be deployed to the API endpoint for real-time scoring. FinTech providers can use Amazon API Gateway and AWS Lambda to make these endpoints available to their consumer-facing digital applications.
4. Batch scoring: The solution also supports batch processing mode through AWS Batch and AWS Fargate, which in this scenario are used for scoring customers of an entire portfolio. The Batch mode is also helpful for use cases that require periodic and continuous monitoring and assessment of the existing portfolio customer credit profile.
5. Customer application: A consumer applies for a credit facility, such as a credit card or BNPL, on the FinTech provider’s digital channels.
6. API response: These requests invoke AWS endpoints, which use the AWS Lambda scoring engine to calculate and deliver the results as real-time insights.
7. Automated credit decision: These insights further drive the customer’s credit facility journey. These models can return insights within the latency requirements of most real-time digital journeys.

Security and Compliance

Data security is a crucial concern for FinTech providers, and sharing data is challenging from a security and compliance perspective. The credit scoring implementation described here helps address these challenges as data remains within the controlled domain and AWS account. This means the FinTech provider retains the customer’s personal financial information. Below is the list of considerations to ensure the security of the data and application.

• Storing raw data in AWS S3 in the encrypted format.
• Storing the processed data and the scoring data in the AWS Redshift data warehouse running in the private network
• Securing the API Gateway endpoint through the access token provided by AWS Cognito.
• Running on-demand and batch process scoring calculations in a virtual private network (AWS VPC)
• Controlling the Inbound and Outbound network traffic through the Virtual Firewall (AWS EC2 security group)
• Storing the secrets in AWS parameter stores

Summary

FinTech providers are now using automated decisions (approve/decline) of credit applications as they realize the benefits it can bring to their business. This is being driven in part by the advent of Open Banking. Many FinTech providers have adopted the cloud in processing such credit decision engines for its resilience and scalability.

This blog covered how FinTech providers can introduce automated credit scoring in the cloud for credit facility applications into their digital journeys. Real-time and batch transform capabilities enable the deployment of a highly scalable, secure, and extensible processing infrastructure with minimal development and operational effort.

What Relevantz can do for you?

Relevantz helps enterprises get more from their data. With our technology acceleration and data engineering services, we build end-to-end data engineering pipelines — covering all six pillars of data engineering — and perform modernization, migration, and maintenance of those pipelines to keep the data flowing as it should.

Do you need help with data engineering solutions in the cloud?