Getting Started
MADI is built as a collection of indepedent components. This enables easier debugging and flexibility in deployments. What suits one organization may not suit another.
The key components are:
- API: FeathersJS in NodeJS
- Database: PostgreSQL / CloudSQL
- Blob Storage: Google Cloud Storage
- Repository Managment: Meta (instead of Git Submodules)
- Parsers: NLM-Ingestor (An Apache Tika-based Document Parser)
- Infrastructure as Code: Terraform
If you are not a developer, setting up the entire codebase for MADI is likely overkill. If you are interested in simply testing the system out locally, you should follow these instructions:
If you are a looking to build a plugin for MADI for your organization, you should proceed to to the plugin instructions:
If you are a developer looking to contribute to MADI or develop a more advanced plugin for MADI , you should proceed to to the advanced instructions:
NOTE: Neither version of MADI currently supports fully local processing. If your organization does not allowing sharing of data with OpenAI, then you should not be using this setup.