Case Study

AI|ML Product Engineering

AI powered automated credit scoring for loan approvals

NutaNXT built a customized AI/ML based customer qualification and scoring model to make real-time decisions for underwriting new loans within seconds.

Client Introduction

A B2C Fintech lending platform company aimed to change the way loans are granted, and wanted to bring together new credit scoring systems for superior customer profiling.

Technologies Used By NutaNXT

  • AI & ML Solutions
  • Predictive Analytics
  • Amazon Web Services

Use Case Applicable To Verticals

  • Banks
  • Insurance
  • Real Estate

Loan

Default Rate

9%

<=

 

$350 M

Loan Volume

Disbursed

Client
Challenge

Banks, Credit cards and other lending financial institutions use credit scores to evaluate potential customers for risk assessment when lending money or providing credit. Traditional credit scoring methods typically need various data such as credit scores, payment history, debt to income ratios, etc. While the largest growth opportunity for credit may be lending to the middle to low-income segments, access to quality credit information on consumers is very limited for this segment, which made it important for the client to build a new automated credit scoring model.

Key challenges with the existing model were:

  • In the absence of quality data, they typically denied credit to a large low income segment of consumers without much consideration of information or specific credit situation.
  • While in some areas, they had overexposure to high risk borrowers, in some ways, our client missed out on business opportunities by not disbursing loans to large numbers of quality borrowers.

Team we built

  • 3
  • Data Scientists
  • 1
  • Cloud Architect
  • 2
  • Data Engineers
  • 1
  • DevOps Engineer

Our Solution

NutaNXT built an AI/ML based credit process and scorecard to determine creditworthiness of new customers factoring large amounts of data enabling the client to run sophisticated models that predict loan default risk and underwrite new loans within seconds.

  • We engineered a variant form of data into a new credit scoring pipeline, and iterated the ML pipeline with distinct algorithms.
  • We formulated feature engineering, calibrated hyperparameters and deployed appropriate ML models for credit scoring.

Technology Stack Used

Loan Default Rate

Reduced To

4%

>=

 

Loan Volume

Increased To

$500 M

Business Impact and Results

Using AI, the loan process was streamlined, which helped improve customer experience, through an advanced credit score.

  • 55% reduction in default rate
  • 42% increase in loan volume
  • Significant Reduction In Portfolio Risk

More Works

From MVP to end-to-end digital products, our culture of intelligent innovation and engineering excellence have solved complex Digital Product Engineering challenges and delivered maximum business ROI for our clients.