Aurora

Aurora

Project Tagline Powered Interactive Robot for Smart Interviews, Real-time Engagement & Cloud Intelligence

What Inspired Us, What We Learned, How We Built It, Challenges We Faced We were inspired by the limitations of current interview and onboarding systems, especially in educational and event-based environments where personalization and real-time responsiveness are lacking. We wanted to build a solution that could not only engage users with smart interactions but also conduct intelligent mock interviews and manage participation data efficiently.

What We Learned Integrating 5G and edge computing for latency-sensitive tasks like gesture detection and voice processing.

Deploying AWS Bedrock to dynamically generate interview questions and analyze voice responses.

Managing multi-component architecture, from resume parsing to real-time scoring and feedback.

How We Built It Developed core functionality using Python and APIs to integrate voice input/output and resume parsing.

Built a mobile companion app using Flutter for live control and feedback.

Set up AWS infrastructure to run real-time services with Bedrock, Polly, Transcribe, and Lambda.

Integrated hardware like Coral TPU and Pi Camera for on-device machine learning.

Challenges Faced Managing real-time data processing within the Raspberry Pi edge layer.

Ensuring smooth handoff between edge and cloud components over 5G.

Developing seamless interaction modules with both pre-trained and real-time AI logic.

Detailed Technical Documentation Aurora is built as a highly scalable platform leveraging AWS GenAI, IoT, and 5G to deliver enterprise-grade automation and interaction capabilities.

Core Technical Elements Raspberry Pi 5 and Coral TPU for edge inference, ensuring sub-second response times.

5G connectivity powers real-time communication between on-device sensors and AWS APIs.

Resume Parser Module: Uses pypdf2 to extract structured data.

Question Generator: Interacts with Amazon Bedrock to create personalized interview questions.

Interview Engine: Includes TTS (Text-to-Speech) using Python libraries, JSON-driven data pipeline.

Scoring Engine: Conducts semantic analysis via Bedrock to provide real-time scoring.

Cloud Services:

Amazon Polly for responses

Amazon Transcribe for inputs

AWS Lambda for function triggers

Amazon S3 / Amazon SES for file/email management

Problem Statement, Solution Overview & Social Impact Problem Manual interview processes are inefficient, inconsistent, and hard to scale. Event registration and engagement systems are often disconnected from real-time data processing or personalization.

Solution Aurora bridges this gap by automating resume parsing, personalized question generation, interview execution, and feedback scoring using cloud-powered intelligence and real-time edge compute.

Potential Impact For Employers & Recruiters: Reduces screening time, introduces objectivity.

For Educators: Automates attendance, provides feedback-driven Q&A.

For Public Engagements: Improves user interaction with language support and gesture recognition.

For Developers: Offers an open platform for extending use cases with plug-and-play AWS modules.

AWS Generative AI Services Used Amazon Bedrock – For personalized question generation and feedback scoring

Amazon Polly – Converts feedback into natural audio responses

Amazon Transcribe – Live speech-to-text for dynamic voice queries

AWS Lambda – Serverless backend for executing custom logic