Sharing Our Passion for Technology
& Continuous Learning
〈  Back to Blog

Case Study: Financial Services Cloud Native Data Lake

Light bulbs with the text Plan Solution Idea

A Fortune 500 financial services partner wanted to transform their future actuarial community, re-engineer and automate manual workflows, and modernize technology tools to gain a competitive position in the market.

We began with a customer-focused approach. Because the end-users consisted of the actuarial community and resident data scientists, we first strove to understand the workflows and existing analytics tooling. We sought to strike a balance in supporting existing tools like PowerBI and SAS and introducing new cloud native tooling like (Sagemaker Notebooks and Amazon QuickSight tied into the data sets). By meeting end users where they were at, while also offering access to new tooling, actuaries could learn the new tooling at their own pace, leading to a higher adoption of the new data lake.

Product

Enterprise Data Lake

We built our partner’s first enterprise cloud data lake and established a framework for actuarial modeling within this new environment that will serve as a blueprint for subsequent analytic model migrations for their offshore partners.

Real-Time Data

Our team deployed a streaming data ingest architecture that delivered data real-time to the lake. With a customer-focused mindset, we gave downstream consumers the option to leverage either batch or stream processing, depending on their needs.

Real-Time Insights

Understanding the dimensions of value for this initiative, we challenged the status quo of batch oriented data sourcing resulting in real-time insights and actionable benefits across the enterprise.

Tech Stack

  • Python
  • PySpark
  • AWS (Lake Formation, Glue, Athena, Sagemaker Notebooks, QuickSight, Cloudformation, Lambda, S3)
  • Debezium
  • DB2 z/OS
  • Docker

Key Results

98%

Automated data for financial compliance reports. Preparation time went from 3 months to 1 day

50+

Framework for actuarial modeling will be the template for 50+ models

12X

Aggregating data from over 12 lines of business for centralized report efficiency

〈  Back to Blog