success story

Customizing NLP summarization with MLOps

Virtusa enhances a leading bank’s NLP summarization solution through an MLOps framework

The Challenge

The client, a leading bank, had several manual and disconnected solutions impeding the automation and scalability of its AI/ML solutions. The client was also using a legacy application, slowing down the development and deployment of its ML solutions. Furthermore, its premature MLOps practices were impacting its productivity and cost of operations. The client also wanted to use the latest AI/ML practices to incorporate fraud detection.

The Solution

Virtusa deployed a hybrid MLOps solution for the client, integrating their on-premises infrastructure with their AWS cloud environment.  

Virtusa delivered the following enhancements:

  • Created a centralized registry to serve AI/ML models on AWS and on-premises.
  • Implemented a Neo4j graph database-based fraud detection solution.
  • Automated the data ingestion process, the endpoint deployment, and continuous training.
  • Integrated the ML pipeline tools with deployment and monitoring tools to create an encompassing ML and Ops pipeline.
  • Implemented data, model, and metadata versioning.
Customizing NLP summarization with MLOps
The Benefit

Post-solution deployment, the client experienced significant improvements in development, deployment, reusability, and reproducibility. 

Benefits include:

  • Reduced development time by 40%.
  • Reduced deployment time by 50%.
  • Significantly reduced integration and testing efforts.
  • Improved repeatability, reproducibility, and predictability.
  • A hybrid MLOps solution integrated with on-premises and cloud environments.
Powering innovation with deep engineering

Learn more about our Artificial Intelligence services

Related content