About this guide

This guide shows you how to build and run your work in Groundzero. You will create projects, write and run ETL jobs, put them on a schedule, work in notebooks, build datasets, train and share models, and connect your own tools.

Each chapter starts with the task and what you get at the end. It then gives numbered steps, a reference for the fields and options, how to check that it worked, and fixes for the messages you may see.

Screenshots show sample data for a company called Acme Retail. Names such as AcmePOSPostgres and daily-sales-rollup are examples, so use your own.

Find your task#

I want to... Go to
Keep my jobs, notebooks and models together Projects
Choose where my code runs and how big it is Runtimes and compute
Move and transform data with Python ETL jobs
Run a job every morning, hour or week Scheduling and run history
Explore data and try ideas interactively Notebooks
Save a query that charts and dashboards can use Datasets
Train, track and share machine learning models Models and MLflow
Drive Groundzero from Codex, Claude Code, Cursor or VS Code Connecting from MCP clients

Before you start#

You need Why
A signed-in Groundzero account with two-step verification set up Groundzero asks for it before the console opens
At least one connection (on the Connections page) Jobs, notebooks and datasets all read from a connection
Permission to use compute To use any runtime except Standard you need a compute size. If the Compute list says Insufficient Permission, ask an administrator for access to compute
Permission to run ETL jobs Without it, running a job is refused

Your administrator grants permissions. When an action is not allowed, Groundzero tells you with a message such as You do not have "…" permission on … or You are not authorized to perform this action. Send that message to your administrator.

Name things the right way#

Most things you create follow one naming rule. The Add or Save button stays greyed out until the name is valid.

Item Allowed characters Rule
Project, job, notebook, dataset Letters, digits, hyphen Start with a letter. Do not end with a hyphen or use two hyphens in a row. No spaces or underscores. Example: daily-sales-rollup
Scheduled task Letters, digits, dot, hyphen, underscore No spaces. The form says Name cannot contain space
Imported model Any text Must be unique. The dialog says Model Name already taken or Model Name is available

A bad project, job or notebook name shows: Name must start with a letter, contain only letters, numbers, and hyphens, and cannot end with or contain consecutive hyphens.

How to read this guide#

  • Bold is a button, menu item or field exactly as it appears on screen.
  • Monospace is code, or a value you type.
  • A Tip is a shortcut or good habit. A Warning is something that can lose work or cost time. A Note is useful background.