Kyros – AI for Actuaries

Next Story

Retina-AI

Using sophisticated predictive analytics, Kyros loyalty forecasting solutions help you manage liability, predict accurate financial returns, and make faster and more informed decisions. I architected and built the multi-petabyte actuarial AI/ML platform using Databricks (Spark), Airflow, and MLFlow on Azure.

Kyros Insights, Chicago, IL

Architected a production-ready AI/ML SaaS platform for actuarial risk analysis for rewards programs for American Express, Hilton Hotels, United Airlines, and more.

Challenged to take an actuarial ML and AI platform from a developer’s POC to a production-ready system.

  • Architected and built a novel ML/AI platform on Azure Databricks (Spark) with MLFlow & Airflow
  • Increased operational efficiency, reducing analysis runs from 2 weeks to 2 days
  • Reduced hardware footprint by 90% – going from 60,000 CPUs to 6,000, resulting in massive cost savings. 

Technologies used: Python, Databricks, Spark, Azure, Postgres, MLFlow, Airflow, Azure Data lakes, Azure DevOps

 

stoneric
2 repositories, 14 followers.

Leave a Reply