Case Studies

ML-DRIVEN INSURANCE TARIFF SELECTION SYSTEM

FINANCE & BANKING
Customer
A large insurance company with $3.1 billion revenue and over 30 million customers.

Challenge
Increasing sales of third-party liability insurance policies for vehicles while maintaining a stable level of risk.

Project Results
A comprehensive automated application processing system has been implemented. It provides customers with personalized rates selected based on predicted accident probability delivered by ML-models which analyze client profile data and insurance history of customers. The system processes more than 10,000 third party liability applications a day and can be scaled to handle applications for other insurance products.

Business Value
Rapid increase of the number of customers due to more competitive personal offers. The ability to quickly make changes to business processes and application processing algorithms, as well as to the applied ML models.

Tech Stack
Camunda, Kubernetes, Jenkins, Jupyter, Mlflow, Feast, Tensorflow, Airflow, Spark Operator, Seldon Core, Git, Java, Apache Kafka