Detailed project case study
EDI-to-API Integration & Middleware Modernisation
A managed data-exchange layer that brought legacy EDI and file-based flows together with modern APIs across internal systems, marketplaces, retailers, carriers and trading partners.
Modernised EDI and file-based trading-partner flows into a hybrid API/EDI middleware layer, improving data exchange, message visibility, exception handling and operational scalability.
The challenge
Retail and fulfilment operations depended on a brittle mix of EDI messages, FTP drops, CSV files, marketplace APIs, carrier feeds and custom mappings. A failed or malformed message could stall an order, distort inventory or leave service teams with no visible explanation.
What I delivered
- Designed a controlled middleware model for routing, transformation, validation, mapping, normalisation and synchronisation.
- Connected API payloads, feeds, files, EDI messages and trading-partner endpoints across cloud and hybrid environments.
- Introduced message traceability, operational monitoring, exception queues and retry handling.
- Supported product, order, inventory, shipment, status and cancellation flows across multiple partners.
- Standardised JSON, XML, CSV, flat-file, SFTP/FTP and EDI patterns behind a consistent operational layer.
- Created a more repeatable route for onboarding and supporting new partner integrations.
Technologies & methods
The delivery combined platform, integration, data and operating practices appropriate to the project boundary.
Business impact
- Improved real-time order and transaction visibility.
- Reduced manual intervention and made failed or rejected messages actionable.
- Improved operational control through traceable exceptions and retry workflows.
- Created a more scalable integration model for retailers, marketplaces, carriers and trading partners.
Why this project matters
This project bridges old-world enterprise integration and modern API-led architecture. That combination matters because critical retail operations rarely modernise in one clean step; the platform had to make both worlds reliable and observable.