PLAIN: A Sovereign Data and AI Platform for Germany

How a data and AI platform was built and how delivery at government scale was enabled.

PLAIN (Platform Analysis and Information Systems) is a data and AI platform for the federal administration. It provides secure and compliant analytics in a sovereign Berlin data centre. The programme office sits in the Foreign Office. Bundesdruckerei develops and operates. Its clients are:

• Federal Ministry for Economic Affairs and Energy
• Federal Ministry of Labour and Social Affairs
• Federal Ministry of Agriculture, Food and Regional Identity
• Federal Ministry of Transport
• Federal Ministry of Health

Challenges are unifying data, analytics, and AI for cross-ministry use while meeting digital sovereignty and accessibility requirements:

• Diverse use cases across peace, conflict, health, climate, and governance
• Strict compliance (BITV 2.0 accessibility, sovereign hosting, procurement constraints)
• Avoiding vendor lock-in with open technologies
• Cross-department collaboration and reuse at scale

I defined and delivered the product backbone:

• Reference architecture: containerised, multi-tenant EaaS with network and identity isolation
• Data layer: data lake + warehouse, data contracts, lineage, and catalog
• Self-service workbenches: Apache Superset for no-code dashboards, GitLab for inner-source, optional ML/Jupyter toolchains
• GPU strategy: virtualised GPUs for elastic AI jobs, quota and scheduling policies
• SDLC/GitOps: CI/CD templates, artefact signing, SBOMs, supply-chain scanning
• Compliance by design: BITV 2.0, GDPR/DPIA patterns, auditable access logs; BSI-aligned controls
• Onboarding playbook: tenant blueprints, landing zones, policy packs, and a reusable “One-for-All” app catalog.

I mapped analyst, data-scientist, and policy personas into concrete flows:

• Secure data onboarding with governed pipelines and PII handling
• No-code dashboards in Superset aligned to ministerial KPIs
• Reusable container app templates for rapid use-case rollout

I enabled cross-department reuse without breaking sovereignty:

• GitLab group standards, code ownership, and MR policies
• Inner-source patterns for dashboards, ETL, and ML components
• Data-sharing agreements and scoped tokens for controlled reuse

I set up low-risk iteration paths:

• Sandbox tenants with synthetic datasets and quota guards
• A pattern library for MLOps, evaluation checklists, and rollback plans

I operationalised trust:

• Role-based access, Just-In-Time approvals, periodic access reviews
• Open-source intake policy, CVE response playbooks, SBOM audits
• Change governance with the PLAIN Program Office and security leads

• Platform live: development since Autumn 2022, Version 1.0 in production since June 2023; Version 2.0 planned from late 2024
• Sovereign, accessible platform implemented per BITV 2.0
• Cross-ministry reuse of data products, models, and workflows
• Recognized with 2nd place in the eGovernment Competition for “Digital Transformation through AI and Modern Infrastructure”

Representative Use Cases

• Monitoring pandemics and political crises with scenario insights
• Optimising funding programs domestically and abroad
• Climate-aware land management planning
• Supply-chain criticality forecasting
• BMZ data products and AI-assisted dashboards