feat: implement MVP for multilingual AI voice-based student re-engagement platform

[JiyaGupta-cs]

Overview

This PR implements the initial MVP for the TAP AI-powered multilingual voice engagement platform designed to re-engage inactive students through conversational voice interactions.

The MVP focuses on:

• outbound AI voice engagement workflows,
• multilingual conversational support,
• student inactivity detection,
• VAPI-based voice orchestration,
• scalable async architecture foundations.

---

#4

Features Implemented

Voice Agent Integration

• Integrated VAPI for realtime conversational voice sessions
• Configured multilingual voice orchestration pipeline
• Added browser-based voice session support for MVP testing

Multilingual Support

• Configured Deepgram Nova 3 transcriber in Multi mode
• Configured Azure Speech multilingual-auto voice provider
• Added multilingual conversational prompting support
• Initial support for:
  • Hindi
  • Marathi
  • Punjabi
  • English

Student Engagement Pipeline

• Implemented inactivity-based student selection logic
• Added nudging engine for identifying inactive learners
• Added personalized conversational variable injection:
  • student_name
  • course_name
  • days_inactive

Backend Infrastructure

• Built FastAPI backend for orchestration
• Added modular service structure
• Implemented VAPI session trigger APIs
• Added health and status endpoints

Queueing Architecture

• Integrated RabbitMQ-based async processing foundation
• Added worker-ready architecture for scalable campaign execution
• Added retry/escalation workflow foundations

Data Layer

• Added dummy student dataset support for MVP testing
• Added optional Frappe LMS integration layer
• Implemented fallback data-loading strategy

Analytics & Logging

• Added call/session logging
• Added engagement event tracking foundations
• Added structured console logging

---

[JiyaGupta-cs]

@manua-glitch

Adding the demo link here for easier review: https://www.tella.tv/video/multilingual-voice-agents-0xte

Multilingual.Voice.Agents.1.mp4

This MVP demonstrates the multilingual AI voice engagement workflow and the current proof of work for the proposed solution. Looking forward to feedback and suggestions for improvement.