Market demands aided by advanced internal and external risk factors are forcing businesses to be more rapid, iterative, reliable and collaborative when executing on transformation initiatives. The client’s focus was to provide a governed framework to allow for continuous model development and deployment.
CLIENT
Large financial services company
REGION
Africa | South Africa
SOLUTION
Automated model deployment
HIGHLIGHTS / OUTCOMES
THE CHALLENGE
The business was experiencing a multitude of issues that included:
- A lack of existing capability to deploy models
- Technical constraints encountered in concluding upfront and ongoing analysis
- A lack of structured and complete data
- An undefined pattern library and ability to govern the model risk lifecycle
BUSINESS OBJECTIVES
- Faster time to deploy models through developer self-service
- Smaller technology footprint by managing only the resources and tools required
- Greater flexibility and capacity in supporting the deployment
SOLUTION IMPLEMENTED
The solution involved implementing a digital model lifecycle leveraging Cloud technology and a suite of applications to manage and monitor model development. This included the development of an execution environment that would serve for model CI/CD, while incorporating a best practice architecture as well as security and data services.
BUSINESS BENEFITS REALISED
The benefits to be gained were numerous, however these were the benefits immediately identified:
- 50% decrease in model development time
- 25% decrease in IT and Business costs
- Easier Python code deployment given the modelled risk management use cases
- Standardisation across the governed model deployment lifecycle
“Deploying and operating containers at scale requires the coordination of infrastructure,
management tooling, and operational practices.”