DataRobot Pulumi Provider for Python

The DataRobot Resource Provider lets you manage DataRobot resources with Pulumi Infrastructure as Code.

Installation

Install the package using pip:

pip install pulumi_datarobot

Configuration

Configure the provider using environment variables or Pulumi config:

# Environment variables
export DATAROBOT_API_TOKEN=your_api_token
export DATAROBOT_ENDPOINT=https://your.datarobot.instance/api/v2

# OR using Pulumi config
pulumi config set datarobot:apikey --secret your_api_token
pulumi config set datarobot:endpoint https://your.datarobot.instance/api/v2

Quick Start

import pulumi
import pulumi_datarobot as dr

# Create a DataRobot use case
use_case = dr.UseCase("my-use-case",
    name="ML Project Use Case",
    description="Created with Pulumi")

# Create a project from a dataset
project = dr.Project("my-project",
    name="Customer Churn Prediction",
    dataset_url="https://s3.amazonaws.com/datarobot-datasets/churn.csv",
    use_case_id=use_case.id)

# Create a deployment
deployment = dr.Deployment("my-deployment",
    project_id=project.id,
    model_id=project.recommended_model_id,
    environment_id="your-prediction-environment-id")

# Export important values
pulumi.export("use_case_id", use_case.id)
pulumi.export("project_id", project.id)
pulumi.export("deployment_id", deployment.id)

Examples

Complete examples are available in the examples directory.

Air-Gapped Environments

For air-gapped deployments:

1. Store state locally

pulumi login --local

2. Install Python dependencies offline

Create wheel directory and download packages:

mkdir wheels
pip wheel pulumi-datarobot -w wheels/
tar cf wheels.tar wheels/

Transfer wheels.tar to your air-gapped system, then install:

tar xf wheels.tar
pip install wheels/* -f wheels/ --no-index

3. Download DataRobot plugin manually

Download the plugin binary from the releases page:

# Replace v0.14.0 with your version, e.g., v0.10.14
pulumi plugin install resource datarobot v0.14.0 --server \
  https://github.com/datarobot-community/pulumi-datarobot/releases/v0.14.0/

4. Skip update checks

export PULUMI_SKIP_UPDATE_CHECK=true

Advanced Usage

Custom Authentication

import pulumi_datarobot as dr

# Using API token credential
api_token = dr.ApiTokenCredential("my-token",
    name="Production API Token",
    api_token="your-secure-token")

# Using basic authentication
basic_auth = dr.BasicCredential("my-basic-auth",
    name="Basic Auth Credential",
    username="your-username",
    password="your-password")

Working with Models

# Register a custom model
custom_model = dr.CustomModel("my-custom-model",
    name="Customer Segmentation Model",
    target_type="Regression",
    target_name="revenue",
    description="Custom model for customer revenue prediction")

# Create a registered model from leaderboard
registered_model = dr.RegisteredModelFromLeaderboard("my-registered-model",
    project_id=project.id,
    model_id="model-id-from-leaderboard",
    name="Best Performing Model")

Resources

• DataRobot Platform
• Pulumi Documentation
• Provider Documentation
• Report Issues

Version

Package version: v0.14.0