Architecture Review and Refactoring for a QA technology company

Overview

Tools & technologies

Challenges

A SaaS platform that offers quality assurance solutions to contact centers, engaged with us to have their platform reviewed and re-architected/re-factored to fix platform's performance issues, functional bugs and to add new features.

Major activities we performed:

  • Analyzed the product through detailed sessions and reviewing existing functionality and architecture.
  • Created missing documentation including Epics/User Stories, Business Flows, Functional Architecture, Logical Architecture, Deployment Architecture, etc.
  • Analyzed code quality and documented code set up guide.
  • Laid down strategy as well as assembled an optimized team to re-architect the architecture, refactor buggy code and optimize cloud (Azure) utilization.
  • Set up ETL pipelines in Azure data factory, Azure functions and logic apps to manage background jobs to ingest and process large volume of data and audios coming from multiple sources."

Results

We analyzed the existing system, code base and APIs, background jobs and database to identify the bottlenecks and documented it. Eventually proposed a solution using cloud services which allowed efficient, error proof and more accurate data processing. AWS Lambda, AWS ECS Far- gate Tasks, Azure Logic Apps, Azure Data factory and Azure functions were used to draft the whole ETL and data processing platform, which is cloud native & serverless. 

We implemented pipelines and serverless achitecture to manage audio & transcription processing. This allowed us to store and process massive audio data along with it's manipulation using AssemblyAl (3rd party machine learning and NLP service).

‍

Tools & technologies

  • Backend APIs and components using .NET framework and.NET core 3.1 using C#
  • Backend deployment - Azure Web Apps, background jobs and multiple Azure VMs running customized legacy services. 
  • Database - Azure SQL Server managed instance, Cosmos DB 
  • ETL pipelines - Azure Data Factory 
  • Process orchestration using Azure Logic Apps 
  • Serverless components - Azure functions and Azure Logic apps
  • Audio processing - FFMPEG running under Azure function
  • ML/AI services - Deepgram and Assembly 

The results

The team successfully migrated the old transcription process, which was using Deepgram 3rd party services and background jobs running on VMs, to Serverless architecture. The new transcription process uses AssemblyAl and provides better transcription, sentiment analysis & deeper insights. The automated pipeline framework is scalable to allow the addion of more clients. Additionally, the team enhanced legacy components by solving bugs, refactoring inefficient code and optimizing complex SQL queries which were causing slowdown of the whole platform

contact us

Get in touch with us

Services you are interested in?
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Head Office:
8 The Green, Suite A,
Dover, DE 19901, US
+1 (315) 5674546