How we enabled a leading loan provider to do customer segmentation and minimize documentation errors.

Business Situation

The client wanted to have a model to identify post loan behavior of defaulters, a credit risk assessment model for better decision making, and solution to minimize documentation errors.

GrowYT Approach

  • We built a team of sector data scientists and data analysts to create insight generation and develop ML & AI based models.
  • We designed and implemented ML based hybrid early warning model predicting EMI default- post loan disbursal.
  • We designed ML based hybrid customer credit risk assessment model – pre loan disbursal.
  • Designed Deep Learning Algorithms for Selfie and Photo match based on facial landmarks.
  • Designed and Deployed Name fuzzy match algorithm.


  • We generated valuable insights based on above models and did segmentation of customer base and evaluated risk associated with each segment.

Benefits & Outcomes

  • Identified distress customers early on which supported client’s collection team.
  • Enabled client to forecast possible loss projections.
  • Automated majority of underwriting process which helped credit underwriters in rational decision making and increased their work efficiency.
  • Algorithms minimized errors in name matching and face recognition to greater extent.

Key Takeaways

  • Models and algorithms to be evolved with changing human behavior.

Events like COVID Pandemic corrupts the data making it futile.