The Solution

Virtusa designed and implemented an AI-powered solution that automates reviewer comments' extraction, classification, and analysis.
The solution extracts all the comments from multiple documents, stores them in a database, and classifies them into four levels using:

Keyword analysis and pattern matching – Identifies common themes and categories.
Classic AI/ML techniques – Enhances accuracy in classification.
Gen AI – Leverages large language models (LLMs) for deeper contextual understanding.

Once classified, the system generates a scorecard based on comment severity and impact. This structured approach enables real-time trending data analytics, helping reviewers identify recurring issues and patterns across multiple reports.

The solution was deployed using a suite of AWS services:

Amazon EC2 for scalable compute capacity.
Amazon RDS for secure and efficient database storage.
Amazon SageMaker to host and optimize the large language models (LLMs) for AI-driven classification.
AWS Web Application Firewall (WAF) to ensure security and compliance.

During development, the team initially planned to use Amazon Bedrock for LLM hosting but later switched to Amazon SageMaker due to its superior data protection and privacy capabilities, ensuring compliance with regulatory requirements.

Services

Industries

Insights

Careers

About