NavigatorAI Setup Guide

NavigatorAI is UF’s AI platform providing access to multiple LLM models including GPT, Llama, Gemini, and Claude. This guide covers setup for the Toolkit API, which you’ll use for course assignments.

Services Overview

Service	Description	Access
NaviGator Chat	Web interface for LLM conversations with custom datasets	Students, Faculty, Staff
NaviGator Toolkit	API access to LLM models	Students, Faculty, Staff
NaviGator Tutor	Personalized learning assistant	Students, Faculty, Staff
NaviGator Notebook	Document assistance with Gemini	Students, Faculty, Staff

For this course, you will primarily use the NaviGator Toolkit for API access.

Getting Started

Step 1: Access the Platform

Go to https://it.ufl.edu/ai/
Click on “NaviGator Toolkit”
Log in with your UF GatorLink credentials

Step 2: Generate an API Key

Navigate to the API Key management section
Click “Generate New API Key”
Copy your API key and store it securely
Never commit your API key to GitHub

Step 3: Set Up Your Environment

Store your API key as an environment variable:

# Add to your shell profile (~/.bashrc, ~/.zshrc, etc.)
export NAVIGATOR_API_KEY="your-api-key-here"

Or create a .env file in your project (add to .gitignore):

NAVIGATOR_API_KEY=your-api-key-here

Python Usage

Install Required Packages

uv add openai python-dotenv

Basic Example

import os

from dotenv import load_dotenv
from openai import OpenAI

# Load environment variables
load_dotenv()

# Initialize client with NavigatorAI endpoint
client = OpenAI(
    api_key=os.getenv("NAVIGATOR_API_KEY"),
    base_url="https://api.navigator.ufl.edu/v1"  # Update with actual endpoint
)

# Make a request
response = client.chat.completions.create(
    model="gpt-4",  # Or other available models
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "What is data engineering?"}
    ]
)

print(response.choices[0].message.content)

Available Models

The Toolkit provides access to multiple models:

Provider	Models
OpenAI	GPT-4, GPT-3.5-turbo
Meta	Llama 2, Llama 3
Google	Gemini
Anthropic	Claude

Check the NavigatorAI documentation for the current list of available models and their identifiers.

Best Practices

Security

Never hardcode API keys in your source code
Add .env to .gitignore to prevent accidental commits
Use environment variables or secrets management

Rate Limits

NavigatorAI has usage limits (check current documentation for specifics)
Implement retry logic with exponential backoff
Cache responses when appropriate

Cost Management

Use smaller models (GPT-3.5-turbo) for development and testing
Reserve larger models (GPT-4) for final runs
Monitor your usage through the Toolkit dashboard

Troubleshooting

Common Issues

Authentication Error (401)

Verify your API key is correct
Check that the environment variable is set: echo $NAVIGATOR_API_KEY
Regenerate your API key if needed

Rate Limit Error (429)

Wait and retry with exponential backoff
Reduce request frequency
Check your usage against limits

Model Not Found

Verify the model name matches available models
Check NavigatorAI documentation for current model list

Getting Help

UFIT Support: Contact UFIT for NavigatorAI platform issues
Course Discord/Canvas: For assignment-specific questions
Office Hours: Discuss API integration challenges

Alternative: Local Models with Ollama

For development and testing, you can use local models via Ollama:

# Install Ollama
brew install ollama  # macOS
# or download from https://ollama.ai

# Pull a model
ollama pull llama2

# Run the model
ollama run llama2

Using Ollama with OpenAI-compatible API:

from openai import OpenAI

client = OpenAI(
    api_key="ollama",
    base_url="http://localhost:11434/v1"
)

response = client.chat.completions.create(
    model="llama2",
    messages=[{"role": "user", "content": "Hello!"}]
)

Resources

back