Getting Started¶

This guide will help you install and configure the Apache Airflow Provider for OVHcloud AI.

Installation¶

Install the provider package using pip or uv:

pip install apache-airflow-provider-ovhcloud-ai

uv init \
uv add apache-airflow-provider-ovhcloud-ai

Configuration¶

Before using the operators, you need to configure Airflow connections for the OVHcloud services you want to use.

AI Endpoints Connection¶

Option 1: Using the Airflow UI¶

1. Navigate to Admin > Connections in the Airflow UI
2. Click + Add a new record
3. Fill in the connection details:

Option 2: Using the Airflow CLI¶

airflow connections add ovh_ai_endpoints_default \
    --conn-type generic \
    --conn-password your-api-token-here

Option 3: Using Environment Variables¶

export AIRFLOW_CONN_OVH_AI_ENDPOINTS_DEFAULT='{"password":"your-api-token-here"}'

AI Training Connection¶

Option 1: Using the Airflow UI¶

1. Navigate to Admin > Connections in the Airflow UI
2. Click + Add a new record
3. Fill in the connection details:

Option 2: Using the Airflow CLI¶

airflow connections add ovh_ai_training_default \
    --conn-type ai_training \
    --conn-login gra \
    --conn-password your-ai-token-here

Option 3: Using Environment Variables¶

export AIRFLOW_CONN_OVH_AI_TRAINING_DEFAULT='{"login":"gra","password":"your-ai-token-here"}'

Getting Your API Tokens¶

AI Endpoints Token¶

1. Go to the OVHcloud Manager
2. Navigate to Public Cloud > AI Endpoints
3. Generate or copy your API token

AI Training Token¶

1. Go to the OVHcloud Manager
2. Navigate to Public Cloud > AI Dashboard > Tokens
3. Generate or copy your AI Training token

Available Regions¶

AI Training Regions¶

Testing Your Connection¶

You can verify your connections are working correctly using the Airflow UI:

1. Go to Admin > Connections
2. Find your connection (e.g., ovh_ai_endpoints_default)
3. Click Test to verify the connection

Or test programmatically:

from apache_airflow_provider_ovhcloud_ai.hooks.ai_endpoints import OVHCloudAIEndpointsHook

hook = OVHCloudAIEndpointsHook()
success, message = hook.test_connection()
print(f"Connection test: {message}")

Your First DAG¶

Here's a complete example DAG that uses OVHcloud AI Endpoints:

from airflow import DAG
from airflow.operators.python import PythonOperator
from apache_airflow_provider_ovhcloud_ai.operators.ai_endpoints import (
    OVHCloudAIEndpointsChatCompletionsOperator
)
from datetime import datetime

def process_response(**context):
    """Process the LLM response from the previous task."""
    ti = context['ti']
    response = ti.xcom_pull(task_ids='generate_text')

    content = response['choices'][0]['message']['content']
    tokens_used = response['usage']['total_tokens']

    print(f"Response: {content}")
    print(f"Tokens used: {tokens_used}")

    return content

with DAG(
    dag_id='my_first_ovhcloud_ai_dag',
    start_date=datetime(2024, 1, 1),
    schedule=None,
    catchup=False,
    tags=['ovhcloud', 'ai'],
) as dag:

    generate = OVHCloudAIEndpointsChatCompletionsOperator(
        task_id='generate_text',
        model='Meta-Llama-3_3-70B-Instruct',
        messages=[
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": "What is Apache Airflow?"}
        ],
        temperature=0.7,
        max_tokens=200,
    )

    process = PythonOperator(
        task_id='process_response',
        python_callable=process_response,
    )

    generate >> process

Next Steps¶

Now that you have the provider installed and configured, explore the specific features:

• AI Endpoints — Learn about chat completions and embeddings
• AI Training — Learn about submitting and managing training jobs
• API Reference — Complete API documentation