Success Story

Operationalizing Risk Scoring Models for one of the largest Global Property & Casualty (P&C) Insurance Firm

About the client

One of the largest global Property & Casualty (P&C) insurance Firm

Challenges

  • Lack of unified source and interface for real time modeling of risk scores and premiums
  • Inability to handle increase in volumes as existing solution was unscalable
  • Delay in handling requests leading to poor customer experience

Approach

  • Devised a solution which was scalable & ensured business continuity with no disruption
  • Built Python based models capable of handling complex scoring rules
  • Ensured seamless data integration by APIfication of models on Azure platform

Solutions

1.Reconfigured the current financial risk scoring models from legacy excel (VB) Macros to optimized Python script to allow real-time batch scoring

2.Built end-to-end containerized model deployment pipeline on MS Azure to enable scalability based on volume

3.Constructed API based framework to allow integration with any front-end solution

4.Enabled data integration with external sources like Dun and Bradstreet, Experian etc., to feed the model to calculate risk scores

5.Built analytical models with 99.7% effective pass rate

Impact Delivered

20%

Reduction in analytical model deployment time

7X

Faster analytics use case development

30 millisecond

Improvement in median latency
for model execution

Tech Stack