ML Workspace Overview
The NATIS ML Workspace provides a fully managed environment for data science — from exploration to model deployment.
On this page
The NATIS ML Workspace integrates Jupyter-compatible notebooks, MLflow experiment tracking, a Feature Store, Model Registry, and one-click deployment into a unified environment — eliminating the need to manage separate MLflow servers, feature pipelines, or serving infrastructure.
ML Workspace Components
Component | Purpose | Technology — | — | — Notebooks | Interactive data exploration and model development | Jupyter-compatible, PySpark, R MLflow Experiments | Track hyperparameters, metrics, artifacts | MLflow 2.x (managed) Feature Store | Reusable, versioned feature definitions for training & serving | NATIS Feature Engine Model Registry | Version, stage, and promote ML models | MLflow Registry Model Serving | Deploy models as REST endpoints or batch jobs | NATIS Model Server AutoML | Automated model selection and hyperparameter tuning | NATIS AutoML Engine
Supported ML Frameworks
NATIS ML clusters come pre-installed with the Databricks Runtime for Machine Learning (NATIS ML Runtime), which includes GPU support, optimized versions of all major frameworks, and NATIS-specific Spark integrations.
- scikit-learn (all estimators)
- XGBoost, LightGBM, CatBoost
- TensorFlow / Keras
- PyTorch
- Hugging Face Transformers
- MLlib (distributed Spark ML)
- Prophet (time series forecasting)
- statsmodels
Was this page helpful?
Thanks for your feedback!