Enhancing Chat Bot Interactions Accuracy for Healthcare Platform

Overview

JashDS enhanced a healthcare platform's chatbot accuracy by 10% by implementing an advanced data ingestion and analysis pipeline, leveraging Azure and Medallion architecture to process 5 GB of daily conversation data and deliver optimized Power BI reports.

About the Client

A platform provider facilitating user-healthcare provider interactions, offering comprehensive reports on engagements across various mediums including calls, IVR, emails, and website interactions.

The Challenge

The client needed to analyze and improve the accuracy of their chatbot interactions while managing large volumes of user conversation data efficiently. They required a seamless experience for reporting chatbot accuracy analysis using the Medallion architecture for ETL processes and Power BI for visualization.

Key Results

Improved chatbot accuracy by 10%, enhancing user experience and satisfaction
Reduced data processing time by 20%, enabling faster decision-making
Increased Power BI report performance by 3x through optimized back-end data processing
Developed a robust data ingestion pipeline, processing 5 GB of daily conversation data

Our Solution

JashDS team implemented a comprehensive solution to address the client's chatbot accuracy analysis needs:

Designed and implemented fact and dimension tables in a star schema to support efficient Power BI reporting
Developed efficient data ingestion pipelines for user conversations with the chatbot, leveraging the Medallion architecture
Implemented daily data fetching and processing from the S3 bucket source
Conducted thorough data cleansing to address all edge cases, ensuring data accuracy and consistency
Stored processed data in the Gold layer for optimized reporting access
Completed all data processing in the back-end, significantly reducing the load on Power BI and enhancing its performance

Technologies Used

AzureS3Medallion architecturePower BIETL (Extract, Transform, Load) processes

Related Case Studies

← Back to All Case Studies

Data Engineering

Azure to AWS SaaS Platform Migration

An EdTech SaaS company migrated its entire Azure-hosted student risk monitoring platform to AWS in 9 weeks — lifting and shifting 17 VMs, 12 PostgreSQL databases, and multiple application services with zero disruption to school district operations. The solution leveraged Terraform, AWS Control Tower, and a fully automated CI/CD pipeline to deliver a scalable, cost-optimized cloud foundation built for rapid expansion.

Data Engineering

Real-Time AI Chatbot Platform’s Lambda to ECS Migration

An AI chatbot startup faced critical Lambda performance issues including 100% memory utilization causing crashes, 7-8 second cold starts,Scalability issues where in multiple concurrent users using this application concurrently faced issues to use the application which includes laginess taking too much time to get the response, application crashing  and completely non-functional WebSocket group chat due to protocol incompatibility between Socket.IO frontend and API Gateway WebSocket backend. Through a 4-week POC engagement, we successfully containerized Lambda functions to ECS Fargate, conducted systematic JMeter load testing up to 1,000 concurrent users, and delivered complete Terraform Infrastructure-as-Code, achieving 94% response time reduction (to sub-1-second), 100% cold start elimination, 0% error rate, and validated linear horizontal scalability while providing all technical documentation and architecture recommendations for production migration decision-making.

Data Engineering

Automated High-Performance VDI for Semiconductor Workforce Development

A US semiconductor workforce development organization needed a secure, scalable cloud desktop platform to provide engineers with access to cost-prohibitive EDA tools but lacked the automated infrastructure to control costs and enforce enterprise identity policies. JashDS delivered a fully automated EC2 lifecycle platform integrating Microsoft Entra ID with Amazon Cognito and NICE DCV, reducing idle compute costs by up to 60% and enforcing 100% MFA compliance across a geo-restricted environment.

Have a similar challenge?

Connect with us