shepard — shepard

shepard (“Storage for HEterogeneous Product And Research Data”) is a multi-database research-data platform developed by DLR’s Center for Lightweight Production Technology (Augsburg). It exposes a single REST API that stores timeseries, files, structured documents, spatial data, and semantic annotations alongside metadata, in support of FAIR research-data management. (Source: aidocs/01-repo-overview.md.)

What it is for

Use case

Research data for an aerospace experiment

Group flight-test artefacts — CAD files, sensor channels, lab-journal entries, derived analyses — under one Collection, navigated as a parent/child DataObject tree with predecessor/successor links for derivations.

Learn more ›

Use case

Time-series ingestion from a measurement campaign

Stream channels into a TimescaleDB-backed timeseries container, then attach the container as a Reference to the DataObject that represents the test run.

Learn more ›

Use case

Cross-DB graph + tabular + spatial in one place

Metadata graph in Neo4j, files and structured documents in MongoDB, timeseries in Postgres+TimescaleDB, optional spatial data in Postgres+PostGIS — addressed through one REST surface, no per-store integrations to maintain.

Learn more ›

Where to next

I want to use shepard
Quickstart with the Python client.
I want to deploy shepard
docker-compose stack, configuration, health endpoints.
I want to extend shepard
Stack, modules, data model, auth model.

At a glance

Java version (backend)

3.27.x

Quarkus version

Persistence stores

FAIR

RDM posture

Aspect	Today
API surface	REST under `/shepard/api`, OpenAPI 3.0 at `/shepard/doc/openapi.json`, Swagger UI at `/shepard/doc/swagger-ui`
Backend	Quarkus 3.27.x on Java 21
Frontend	Nuxt 3 + Vue 3 + Vuetify 3 (separate Docker image)
Persistence	Neo4j 5.24 + MongoDB 8.0 + Postgres+TimescaleDB; optional Postgres+PostGIS
Auth	External OIDC (Keycloak typical) + secondary `X-API-KEY` header
Deployment	docker-compose, on-premises (no cloud assumed)

Snapshot date: 2026-05-07.