Developing a Conversational Test Creation Tool for Elephant Scale

Client

Elephant Scale

Location

Delaware

Industry

Ed-Tech

Services & Tech

AWS Transcribe, AWS Bedrock, AWS SageMaker, AWS S3, RDS, Lambda, and API Gateway

Project Overview

Elephant Scale, a leader in personalized learning solutions, aimed to develop an end-to end conversational test creation tool. This tool would automatically transcribe audio recordings of online classes, generate quiz questions, and manage backend data pipelines. By leveraging AWS AI services, Elephant Scale sought to enhance personalized learning experiences for students. Avahi, an advanced tier AWS partner, was engaged to implement this innovative solution using AWS services such as Amazon Bedrock, AWS Transcribe, SageMaker, S3, RDS, Lambda, and API Gateway.

The
Problem

Elephant Scale faced several challenges in creating an automated test generation tool:

  • Real-Time Processing: Capturing and transcribing audio recordings of online classes in near real-time.
  • Data Management: Efficiently managing and processing large volumes of educational content and audio data.
  • Accuracy: Ensuring high accuracy in transcriptions and quiz question generation.
  • Scalability: Building an infrastructure capable of handling multiple concurrent users and large datasets.
  • Integration: Seamlessly integrating the new solution with existing educational platforms and data sources.

Why AWS

Key Deliverables

  • POC Development: Building the initial POC for an end-to-end automated testing tool on AWS.
  • Real-Time Transcription: Developing a system to generate transcripts from audio recordings of online classes in near real-time.
  • ETL Processes: Performing ETL on raw data to generate quizzes after each class and store the output text files in Amazon S3.
  • Backend Pipeline Management: Creating an end-to-end pipeline to pull data from whitelisted third-party sources.
  • Post-Processing Jobs: Running post-processing jobs based on confidence thresholds selected by Elephant Scale.
  • Validation: Ensuring accuracy, bias mitigation, and scalability using industry standard tools.
  • Qualitative Review: An analysis report based on manual review to identify common issues and patterns.

Solution Architecture

The solution architecture leveraged various AWS services to implement the deliverables and ensure a robust, scalable platform.

Data Acquisition and Storage

  • Amazon S3: Used for storing input documents, audio recordings, course content, and processed data. S3’s scalability and durability ensured efficient handling of large volumes of data.

Event-Driven Processing

  • AWS Lambda: Deployed to run event-driven functions that process data and manage various stages of the pipeline. Lambda functions were used to fetch documents and class recordings, create embeddings, and update/store outputs.

Machine Learning and AI

  • AWS Transcribe: Employed for automatic speech recognition, enabling real-time transcription of audio recordings.
  • AWS SageMaker: Utilized for developing, training, and deploying machine learning models that generate quiz questions and manage the transcription process.
  • Amazon Bedrock: Used for scalable managed Generative AI and foundational model services, supporting the overall AI infrastructure.

API Management

  • Amazon API Gateway: Provided a managed service to create, publish, maintain, and secure APIs. This facilitated seamless communication between the front-end applications and backend services, allowing easy access to transcribed data and generated quizzes.

Monitoring and Logging

  • AWS CloudWatch: Implemented for comprehensive monitoring and logging. CloudWatch provided real-time insights into system performance, user interactions, and the health of the deployed models.

Solution

Avahi proposed a comprehensive solution to address these challenges, focusing on developing a proof of concept (POC) for the automated test generation tool. The project was executed in four phases: Discovery & Planning, Design & Development, Validation & QA, and Executive Presentation & Handoff.

We highly recommend Avahi as a reliable and innovative technology partner. Their expertise in cutting-edge technologies was instrumental in building our Proof of Concept (PoC) and developing our Minimum Viable Product (MVP). Avahi consistently delivered high-quality solutions on time while maintaining a collaborative, responsive approach. They went beyond expectations by identifying opportunities for enhancement, ensuring scalability and compliance for our law enforcement-focused products. Avahi is the clear choice if you need a tech partner with industry knowledge, professionalism, and a commitment to innovation.

Brandon Puhlman

Founder, Bravo Foxtrot

Ready to Transform Your Business with AI?

Book Your Free Ignition AI Workshop

Let’s explore your high-impact AI opportunities together in a complimentary half-day session

View Our Case Studies

See how we’ve delivered measurable results for businesses like yours