Husky Navigator

Husky Navigator is an AI-powered assistant for Northeastern University's Silicon Valley campus. It provides information about courses, faculty, academic calendars, degree requirements, and class schedules through a conversational interface.

Overview

This system leverages a Retrieval-Augmented Generation (RAG) approach with the open-source Llama 3 model to provide accurate, university-specific information. It includes:

• Six specialized tools for different query types (course_search, faculty_search, academic_calendar, degree_requirements, course_schedule, northeastern_knowledge_base)
• LLM-based tool selection for optimal query routing
• Document-specific chunking strategies for improved retrieval
• Fallback RAG mechanism for robust error handling
• Streamlit web interface for easy interaction

Setup Instructions (Google Cloud VM)

1. Create a Google Cloud VM

1. Go to Google Cloud Console and create a new VM instance
2. Select the following specifications:
   • Machine type: Choose a GPU-enabled instance (NVIDIA L4)
   • Operating system: Deep Learning on Linux
   • Version: Deep Learning VM with CUDA 12.3 M129
   • Boot disk size: 50 GB
   • Firewall settings:
     • Allow HTTP traffic
     • Allow HTTPS traffic
     • Allow Load Balancer Health Checks

2. Connect to your VM

Once your VM is running, connect via SSH

3. Install and Run Husky Navigator

Run the following commands:

# Clone the repository
git clone https://github.com/Ren-97/Husky_Navigator.git

# Change directory
cd Husky_Navigator

# Make the entrypoint script executable
chmod +x entrypoint.sh

# Build the Docker container
sudo bash ./docker-startup build

# Deploy with GPU support
sudo bash ./docker-startup deploy-gpu

4. Access the Web Interface

Open your browser and navigate to:

http://[EXTERNAL_IP]:8501

Replace [EXTERNAL_IP] with your VM's external IP address.

Usage

Once you access the web interface, you can:

• Ask questions about courses, faculty, academic calendar, degree requirements, and more
• Use the toggle to control conversation memory and summary mode (Reset it before you want to memorize conversation)
• Reset the conversation when needed

Example questions:

• "Tell me about DS 5110."
• "Who will teach NLP in fall 2025?"
• "Who is Karl Ni?"
• "What is the first day of summer break?"

Example Conversation with memory:

• "Reset"
• "Who is Karl Ni?"
• "Will he teach NLP in fall 2025?"

Technical Details

• Frontend: Streamlit
• LLM: Llama 3
• Embedding Model: nomic-embed-text
• Vector Database: ChromaDB
• Framework: LangChain

Extra Note

If Husky Navigator didn't correctly select the tool, you can add the tool you want after query. For example, query + 'use [tool]'