Deployment Options: Library vs. Microservice

Data Designer is available as both an open-source library and a NeMo Microservice. This guide helps you choose the right deployment option for your use case.

Deployment Architectures at a Glance

Data Designer supports three main deployment patterns:

Quick Comparison

📦 When to Use the Open-Source Library

The library is the right choice for most users. Choose it if you:

You Have Access to LLMs

You have API keys or endpoints for LLM inference:

• Cloud APIs: NVIDIA API Catalog (build.nvidia.com), OpenAI, Azure OpenAI, Anthropic
• Self-hosted: vLLM, TGI, TensorRT-LLM, or any OpenAI-compatible server
• Enterprise gateways: Centralized LLM gateway with RBAC, rate limiting, or other enterprise features

1
from data_designer.interface import DataDesigner
2
from data_designer.config import ModelConfig
3
4
# Use any OpenAI-compatible endpoint
5
model = ModelConfig(
6
    alias="my-model",
7
    model="nvidia/nemotron-3-nano-30b-a3b",
8
    provider="nvidia",  # or "openai", or a custom ModelProvider
9
)
10
11
dd = DataDesigner()
12
# Your code controls the full workflow

You Need Maximum Flexibility

• Custom plugins: Extend Data Designer with custom column generators, validators, or processors
• Local development: Rapid iteration with immediate feedback
• Integration: Embed Data Designer into existing Python pipelines or notebooks
• Experimentation: Research workflows with custom models or configurations

You Already Have Enterprise LLM Infrastructure

1
from data_designer.config import ModelConfig, ModelProvider
2
3
# Define your enterprise gateway as a provider
4
enterprise_provider = ModelProvider(
5
    name="enterprise-gateway",
6
    endpoint="https://llm-gateway.yourcompany.com/v1",
7
    api_key="ENTERPRISE_LLM_KEY",  # Environment variable name (uppercase) or actual key
8
)
9
10
# Use the provider in your model config
11
model = ModelConfig(
12
    alias="enterprise-llm",
13
    model="gpt-4",
14
    provider="enterprise-gateway",  # References the provider above
15
)

☁️ When to Use the Microservice

The NeMo Microservice exposes Data Designer’s preview and create methods as REST API endpoints. Choose it if you:

You’re Using the NeMo Microservices Platform

The primary value of the microservice is integration with other NeMo Microservices:

• NeMo Inference Microservices (NIMs): Seamless integration with NVIDIA’s optimized inference endpoints
• NeMo Customizer: Generate synthetic data for model fine-tuning workflows
• NeMo Evaluator: Create evaluation datasets alongside model assessment
• Unified deployment: Single platform for your entire AI pipeline

You Want to Expose SDG as a Team Service

If you need to provide synthetic data generation as a shared service:

• Multi-tenant access: Multiple teams submit generation jobs via API
• Job management: Queue, monitor, and manage generation jobs centrally
• Resource sharing: Shared infrastructure for SDG workloads

When users can submit configs containing Jinja templates to a shared engine, template rendering becomes a remote code execution concern and part of your security boundary. See Security for guidance on when to keep the default JinjaRenderingEngine.SECURE mode.

🧭 Decision Flowchart

                    ┌─────────────────────────┐
                    │ Are you using the NeMo  │
                    │ Microservices platform? │
                    └───────────┬─────────────┘
                                │
                    ┌───────────┴───────────┐
                    ▼                       ▼
                   YES                      NO
                    │                       │
                    ▼                       ▼
        ┌───────────────────┐   ┌───────────────────────────┐
        │ Use Microservice  │   │ Do you need to expose SDG │
        │                   │   │ as a shared REST service? │
        │ Integrates with   │   └─────────────┬─────────────┘
        │ NIMs, Customizer, │                 │
        │ Evaluator         │     ┌───────────┴───────────┐
        └───────────────────┘     ▼                       ▼
                                 YES                      NO
                                  │                       │
                                  ▼                       ▼
                      ┌─────────────────────┐   ┌─────────────────┐
                      │ Consider if the     │   │ Use the Library │
                      │ overhead is worth   │   │                 │
                      │ it vs. library +    │   │ Most flexible   │
                      │ enterprise gateway  │   │ option for      │
                      └─────────────────────┘   │ direct use      │
                                                └─────────────────┘

Learn More

• Library: Continue with this documentation
• Microservice: See the NeMo Data Designer Microservice documentation
• Security model: See Security