Data & ML

Production ML Training Pipeline

S3 → Spark feature engineering → Feast → SageMaker → MLflow → A/B deployment

AI Prompt

“Draw a production ML training pipeline: raw data in S3 → feature engineering (Spark on EMR) → feature store (Feast) → model training (SageMaker) → experiment tracking (MLflow) → model evaluation → model registry → A/B deployment with shadow mode.”

Try this in AIDrawIO

Free — no credit card required

Refine it with follow-up prompts

After generating the base diagram, use these prompts to iterate and add detail — the same way a real architect would refine a whiteboard sketch.

  • 1

    “Add Great Expectations data validation before feature engineering”

    Try this follow-up
  • 2

    “Show feedback loop from production predictions back into training”

    Try this follow-up
  • 3

    “Add data drift detection and model performance degradation alerts”

    Try this follow-up

How AIDrawIO generates this diagram

  1. 1.You paste the prompt above into the chat input.
  2. 2.AIDrawIO sends it to your chosen AI model (GPT-5, Claude, or Gemini).
  3. 3.The model returns draw.io-compatible XML — rendered instantly in the canvas.
  4. 4.Export as SVG, PNG, or XML. Edit any element manually or with follow-up prompts.
All examples