We validate the success of a model against the domain-driven use cases daily by comparing it against the ML problem (framed) and the system level resources used to support the phases of ML workloads. Our end-to-end solution building life cycle contains the following set of activities with automation applied at every stage:

Business goal identification
ML problem framing
Data collection and integration
Data preparation
Data visualization and analytics
Feature engineering
Model training
Faster AI (Virtusa and Stanford AI Laboratory collaboration)
Model evaluation
Business evaluation
Model deployment
Transfer learning
Federated learning
MLOps

We use Amazon S3 as a data lake to hold large data volumes such as electronic medical records (EMRs), claims, etc. An Amazon SageMaker notebook to run against the P3 instance types and G4 EC2 instances. Data cleansing, processing, discovery, and feature engineering get completed in both the SageMaker training instance and G4 EC2 instances. We also use Amazon SageMaker to train custom models using the transformed data, leveraging the distributed training capability, and creating an Auto Scaling API endpoint. The API endpoint makes batches and real-time inferences.

Services

Industries

Insights

Careers

About

Services

Industries

Insights

Careers

About