Elytra
Revolutionizing student counseling with an AI-powered video bot for personalized international education guidance
Created on 25th November 2024
•
Elytra
Revolutionizing student counseling with an AI-powered video bot for personalized international education guidance
The problem Elytra solves
Challenge
Navigating international education can be overwhelming for students due to:
- The sheer volume of programs available.
- Complex and diverse eligibility criteria.
- Limited access to personalized and timely guidance.
Traditional counseling methods, such as static FAQs or general forms, fail to deliver the level of personalization and immediacy required, leaving students feeling lost and underprepared.
Solution
Elytra solves this by offering:
- Human-like Video Bot:
- Engages students with voice and text communication.
- Provides real-time responses and personalized counseling.
- AI-Driven Personalization:
- Recommends programs tailored to student profiles and eligibility.
- Offers context-aware conversations using Natural Language Processing (NLP).
Benefits
- Efficiency: Immediate and automated responses save students’ time.
- Personalized Recommendations: AI ensures accuracy and relevance in suggestions.
- Enhanced Engagement: Human-like interactions foster trust and satisfaction.
Comparison to Existing Solutions
Unlike traditional methods, Elytra delivers dynamic and interactive counseling sessions tailored to the unique needs of each student, providing a superior experience for international education seekers.
Challenges we ran into
1. Speech-to-Speech API Integration
- Obstacle: Integrating the Agaro Speech-to-Speech API for real-time, seamless conversion of user input into dynamic conversational responses.
- Solution: Deployed the backend handling Agaro API, on AWS EC2 service
2. Animated Avatar Integration
- Obstacle: Creating a human-like animated avatar capable of synchronized lip movements and realistic expressions.
- Solution: Combined animation libraries with real-time phoneme mapping to achieve expressive animations during interactions
3. LLM API Integration
- Obstacle: Implementing a scalable, low-latency API layer for integration with Large Language Models (LLMs) to enable real-time context-aware query resolution.
- Solution: Developed an API management system that efficiently routes requests to the LLM, caches frequent queries, and maintains conversation state for improved user experience.
4. Program Recommendation System with Pinecone
- Obstacle: Building a recommendation system capable of dynamically filtering programs based on complex eligibility criteria and user profiles.
- Solution:
- Integrated Pinecone API for vector-based search and matching.
- Combined with a scoring algorithm to rank the most relevant programs tailored to student needs.
Discussion
Builders also viewed
See more projects on Devfolio