The 'Data Science Virtual Machine (DSVM)' is a 'Ubuntu 20.04' VM that has several popular tools for data exploration, analysis, modeling & development pre installed.

Highlights:

Operating System, Drivers and other base components

• "Ubuntu 20.04" image from Azure Marketplace
• Nvidia drivers, CUDA Toolkit, cuDNN (when GPU machines are used)
• Docker
• Anaconda ("conda")
• Git

Authoring Tools

• Visual Studio Code
• PyCharm Community Edition
• Jupyter, Jupyter Lab

ML Framework

• PyTorch, TensorFlow, scikit-learn
• pyspark
• dask
• Vowpal Wabbit

Other Notable Components

• Azure CLI
• Azure ML SDK for Python
• Azure Storage Explorer

Users can access the DSVM via Remote Desktop, SSH or browser (eg: Jupyter Hub). Optionally, the VM can be placed in a corporate network.

Users have full access to the DSVM. If needed, configurations can be adjusted, and additional frameworks can be installed like with any other virtual machine. The image provided here is a static VM image. Maintenance and protection against vulnerabilities of provisioned DSVMs is in the customer's responsibility.

There is also a DSVM on Windows.

For more details, see the Data Science Virtual Machine documentation.