Multi-Project Outerbounds Repository

This repository demonstrates how to organize multiple Outerbounds projects in a single repository using the obproject_multi.toml configuration file.

Repository Structure

.
├── obproject_multi.toml      # Multi-project manifest
├── ml-infra/                 # ML Infrastructure project
│   ├── obproject.toml
│   └── flows/
│       └── test/
│           └── flow.py
└── services/                 # MLOps Services project
    ├── obproject.toml
    └── deployments/
        ├── api/
        │   ├── main.py
        │   └── config.yml
        └── bi-dashboard/
            ├── app.py
            └── config.yml

Projects

1. ML Infrastructure (ml-infra)

Contains Metaflow workflows for machine learning pipelines and data processing.

2. MLOps Services (services)

Contains deployed applications:

• API: FastAPI service for model serving
• BI Dashboard: Streamlit dashboard for business intelligence

Deployment

Deploy all projects:

obproject-deploy

Deploy a specific project:

obproject-deploy --project ml_infra
obproject-deploy --project mlops_services

CI/CD

The repository includes CI/CD configurations for:

• GitHub Actions (.github/workflows/deploy.yml)
• Azure DevOps (azure-pipelines.yml)

Environment Variables

For CI/CD deployments, set:

• CONTINUE_ON_ERROR=1 to continue deploying other projects if one fails
• SPEC_ONLY=1 to only generate project specifications without deploying