Machine Learning Lifecycle Management
Banking & Financial ServicesBusiness Impacts
Reduction in human intervention
5 - 2 days
Reduction in provisioning time
Improvement in customer satisfaction
Customer Key Facts
- Location : North America
- Industry : Financial Services
Problem Context
The customer, an international financial services company, provides a diverse range of wealth accumulation and protection products and services. Their existing process for data preparation, model selection, and testing was manually driven and riddled with inefficiencies; taking up to five days in resource provisioning.
Challenges
- Integrating with third party solutions to automate the process of Amazon SageMaker resource deployment
- Automating the process from data preparation to model selection and testing on Amazon SageMaker
Technologies Used
Amazon SageMaker
AWS CloudFormation
Amazon Simple Storage Service
Automating Machine Learning Lifecycle Management for Increased Efficiency and Cost Savings
The customer sought to automate the process of providing Amazon SageMaker resources to end users; better enabling its developers and data scientists to create, train, and deploy machine-learning models in the cloud.
Solution
Quantiphi developed predefined cloud formation templates (CFTs) that deployed the necessary Amazon SageMaker resources based on a data scientist’s requirements. The solution includes a model leaderboard for performance evaluation, and uses a combination of custom-built pieces as well as the Autopilot, Automatic Model Tuning, Endpoints, and Model Monitor features of Amazon SageMaker.
Result
- Automated model training
- Reduced manual intervention
- Cost reduction
