Tharindu Ranathunga
Back to projects

CORDS: Collaborative Data Space for Manufacturing

Jan 2023 - Jan 2026

Federated learning platform on International Data Space protocol for self-sovereign, cross-party model orchestration in manufacturing.

Problem Statement

  • Manufacturing partners need secure model collaboration without surrendering data control.
  • ML workflows require semantic interoperability to scale across organizations.
  • Federated orchestration needs trustworthy execution and governance alignment.

What I Led

  • Led design and implementation of a minimal viable data space infrastructure and semantic library for ML workflows.
  • Designed and implemented decentralized federated ML services over data space protocol components.
  • Defined technical pathways for governance-aware model lifecycle operations.

Deliverables

  • Data space software stack and semantic assets
  • Federated ML orchestration services
  • Manufacturing-oriented model exchange workflows

Architecture Placeholder

Diagram slot for system architecture and data flow.
PythonJavaEESPARQLApache CamelOPCUADockerKubernetesPyTorchFlowerMLflowKafka

Industry Relevance

Demonstrates how manufacturing ecosystems can operationalize AI collaboration while preserving sovereignty, governance, and interoperability.

Problem framing

Cross-party AI collaboration fails when data sovereignty, interoperability, and trust are treated separately. CORDS integrated these concerns into one architectural track.

Delivery pattern

  • Data space-first architecture
  • Semantics-driven interoperability
  • Federated services with enforceable policy controls

Why it matters

The pattern enables industrial partners to move from isolated pilots toward governed, production-oriented ML collaboration.