Projects

A project in AIchor links a VCS repository (GitHub, GitLab, Bitbucket, or Azure DevOps) to a compute engine and allows users to run workloads.

What a project provides

When a project is created, AIchor provisions the following:

• A webhook integration with the linked repository so that commits prefixed with EXP or exp trigger experiments automatically
• An artifact registry for storing Docker images built during experiment runs
• Object storage buckets backed by the engine's cloud provider to store experiment inputs/datasets, experiment outputs and experiment logs

For Kubernetes engines:

• A Kubernetes namespace on the engine where workloads are scheduled
• Possibility to create pvcs to share data between experiments and avoiding downloading/uploading from/to the internet

Experiments can also be submitted without the webhook by using the AIchor CLI.

Project dashboard

The Projects section is accessible from the left navigation menu. The dashboard shows the full list of projects in the organisation, with each entry displaying:

• Project name
• Creation date
• Link to the associated VCS repository
• Associated engine
• Project status

The project list can be filtered by name, namespace and engine.