SummitChatbot

An AI-powered conversational onboarding system designed for student-athletes, featuring multilingual support, fine-tuned language models, and intelligent data extraction.

Overview

SummitChatbot is a comprehensive chatbot solution that guides users through an onboarding process, collecting personal, academic, athletic, and career information through natural conversation. The system combines multiple AI technologies including fine-tuned language models, RAG (Retrieval-Augmented Generation), and rule-based extraction to provide intelligent, context-aware responses.

Architecture

The system consists of several interconnected services:

• Onboarding API: Main conversational interface with intelligent field extraction
• RAG Server: Fine-tuned language model service with knowledge base integration
• Rasa Framework: Intent recognition and dialogue management
• Quality Monitor: Response quality tracking and analytics
• Multilingual Support: Multi-language conversation capabilities

Key Features

• Conversational Onboarding: Natural language data collection for student profiles
• Intelligent Extraction: Automatic parsing of user input into structured data
• Multilingual Support: Conversations in multiple languages
• Fine-tuned Models: Custom-trained language models for domain-specific responses
• Quality Monitoring: Real-time response quality assessment
• Docker Deployment: Containerized services with Docker Compose
• Knowledge Base Integration: RAG-powered responses using curated knowledge

Technology Stack

• Backend: FastAPI, Python
• AI/ML: Hugging Face Transformers, RAG, Fine-tuning
• NLP: Rasa, Sentence Transformers
• Deployment: Docker, Docker Swarm
• Monitoring: Custom quality assessment tools

Quick Start

1. Prerequisites: Docker, NVIDIA GPU (for model inference)
2. Setup: Configure API keys in secrets/ directory
3. Deploy: Run docker-compose up to start all services
4. Access: API available at http://localhost:8000

Project Structure

summit/
├── service/           # Main onboarding API
├── fine_tuning/      # Model training and RAG services
├── rasa/             # Rasa chatbot framework
├── scripts/          # Deployment and testing scripts
└── secrets/          # API keys and configuration

Services

• Port 8000: Onboarding API (main interface)
• Port 5055: Rasa Action Server
• Internal: RAG Server (model inference)

Use Cases

• Student-athlete onboarding and profile creation
• Academic and athletic information collection
• Career experience documentation
• Multilingual user support
• Automated data validation and structuring

Development

The project supports both local development and production deployment with Docker Swarm, featuring automated CI/CD pipelines and secure secret management.