CreditVidya Extends the Loan Market to Millions of Financially Excluded Indians with AWS

creditvidya

Financial Inclusion for Millions

CreditVidya is a startup headquartered in India whose underwriting technology is opening the country’s loans market to over 250 million financially excluded citizens. Traditionally, financial institutions have been unwilling to lend to these citizens—typically with a daily household income of between $2 and $10—because they lack collateral and a credit history. Furthermore, processing the loans, which averages about $290 per applicant, has been too costly. CreditVidya’s technology is reducing the cost of processing loans from about $2 to less than one cent while overcoming a lack of credit history or collateral by leveraging loan applicants’ digital footprints to measure their creditworthiness.

To determine creditworthiness, CreditVidya’s artificial intelligence (AI) and machine learning (ML) platform—Medhas—leverages payment data, financial behavioral data, and device data stored on smartphones to help determine loan applicants’ ability and intent to repay loans. Medhas platform’s AI technology continuously optimizes the parameters for scoring applicants’ creditworthiness based on a growing store of loan applications. Since working with CreditVidya, lending partners—which include 55 leading banks and non-banking financial institutions in India—have seen loan approval rates increase by 25 percent and delinquency rates decrease by 33 percent.

Srikanth Gaddam, VP of IT & Security at CreditVidya, says, “Our decision-strategy engine optimizes the delinquency rates for each lender according to their appetite for risk. The information we assess is far more comprehensive than traditional application and bureau-based scorecards, so lenders can make richer policy decisions on customers, reducing overall portfolio delinquencies.”

“With AWS, CreditVidya is helping millions of people gain access to India’s formal loans market at lower interest rates.”

Srikanth Gaddam, VP of IT & Security, CreditVidya

  • About CreditVidya

  • CreditVidya enables access to affordable institutional credit to financially excluded citizens. It does this by running artificial intelligence and machine-learning models on traditional and alternative data for superior credit-risk assessment.

  • Benefits
    • Complies with banking and finance requirements with a secure cloud platform
    • Scales to support 10 times more daily loan applications
    • Builds data lakes holding 250–500 TB of data to analyze loan performance
    • Enables global expansion with a cost-effective platform
  • AWS Services Used

A Secure Platform Born in the Cloud

CreditVidya has been running the Medhas platform on the Amazon Web Services (AWS) Cloud since day one. The AWS pay-as-you-grow model of cloud computing meant CreditVidya avoided any large upfront capital costs for building the platform. In addition, the security of the AWS infrastructure and services such as Amazon GuardDuty, a threat-detection service that continuously monitors for malicious activity, made AWS more attractive to CreditVidya than other cloud service providers. Gaddam says, “Amazon GuardDuty provided a one-stop solution to protect us against a range of threats such as malware.” CreditVidya also leverages AWS CloudTrail, Amazon CloudWatch, AWS Inspector, Amazon Virtual Private Cloud (Amazon VPC) flow logs, and AWS Identity and Access Management (IAM). “This suite of security services influenced our decision to use AWS because they help us build trust with our clients,” says Gaddam.

CreditVidya also found that AWS has audit-friendly service features to meet International Organization for Standardization (ISO), Security Operations Center (SOC), and Payment Card Industry Data Security Standard (PCI-DSS) requirements. In addition, the AWS compliance framework met the needs of CreditVidya’s banking and finance clients. Gaddam says, “We found potential customers had an increased level of confidence in our service, knowing that it ran on AWS.”

About 250 Terabytes of Digital Footprints

CreditVidya’s Medhas platform collects the digital footprints of loan applicants via a software development kit (SDK) installed in lenders’ mobile apps. Applicants download the app and apply for a loan, giving CreditVidya permission to collect certain data from the applicants’ device. The footprint data is ingested by the Medhas platform via APIs.

The company’s Java-based application running on Amazon Elastic Compute Cloud (Amazon EC2) instances decrypts the ingested data and runs ML algorithms to compute the variables that will determine applicants’ credit score. The Amazon EC2 instances sit inside a parallel-processing Hadoop framework, which is dynamically scaled by Amazon Elastic MapReduce (Amazon EMR). With AWS Auto Scaling and the use of Amazon EC2 Spot Instances, CreditVidya maximizes the efficient running of the platform while having the instant scalability to handle traffic spikes when partners launch promotional campaigns for their loan services.

Today, the Medhas platform has about 250 TB of structured and unstructured data held in Amazon Relational Database Service (Amazon RDS), Amazon DynamoDB, and Amazon Simple Storage Service (Amazon S3). Data is stored in Amazon S3, with metadata from that data set held in Amazon DynamoDB. When querying large datasets, the business first queries the Amazon DynamoDB table and gets a list of objects to be fetched from Amazon S3.

In the last three months, CreditVidya has started using Amazon S3 and AWS Glue to build data lakes. It plans to use the data lakes to analyze the performance of its partners’ loans by demographics, occupation, or geographical location. The startup expects its data lake to hold approximately 500 TB of data over the next two years.

Ingesting Data from 100,000 Daily Loan Applications

The Medhas platform has gone from ingesting digital-footprint data relating to 10,000 daily loan applications to 100,000 a day in the last 12 months as the number of CreditVidya loan partners has grown to over 55. The startup can easily scale the platform as the number of applications increase with AWS Auto Scaling, “We’ve helped with the processing of more than 25 million loan applications for our lending partners on AWS,” says Gaddam.

CreditVidya plans to go live soon with Amazon Rekognition, which adds facial recognition to applications. Comments Gaddam, “We plan to use Amazon Rekognition to complete our electronic “know your customer” processes. We will compare users’ uploaded identity cards and selfies to ensure that applicants are uploading their own identity cards.”

Enabling Lower Rates of Interest

CreditVidya is enabling 80 percent of the Indian population to apply for loans of $150 to $3,000. Gaddam comments, “Many people in India are forced to go to expensive, informal sources of credit because they don’t have security for their loans. Because our delinquency rates are often better than the industry average, lenders peg their interest rates for the loans we facilitate to the ones available to regular customers and businesses. These rates start at about 14 percent. With AWS, CreditVidya is helping millions of people gain access to India’s formal loans market at lower interest rates.”

Extending Credit to 3 Billion People

CreditVidya is experiencing an average of 55 percent year-on-year business growth with AWS. The company hopes to increase its rate of expansion by launching its services in different parts of the world, including Southeast Asia. “There are approximately 3 billion people around the world with no access to formal loans,” says Gaddam. “With AWS, we hope to address this problem by partnering with banks and other financial institutions, so they have an effective way of providing credit to this people at a low cost.”

Learn More

To learn more, visit https://aws.amazon.com/machine-learning/