Customer case
Radiometer Translation System is a custom-developed translation management system that handles translation of IDML, DOCX and PPTX documents to multiple languages in one unified platform. The solution combines machine translation, translation memory, glossaries and a dedicated validator workspace so Radiometer can ensure consistent and quality-assured translations – also for non-Latin languages.
Radiometer is a global company in medical technology equipment and diagnostic solutions, with particular focus on blood gas analysis, point-of-care diagnostics and critical patient monitoring.
The company develops and delivers advanced solutions to hospitals and laboratories worldwide, where documentation, quality and regulatory requirements place high demands on linguistic precision and consistency across many markets and languages.
Visit radiometer.comRadiometer Translation System is a functional, server-based translation system that replaces an earlier desktop-based solution for handling Radiometer's multilingual technical documents. The system automates the chain from upload of IDML, ZIP, DOCX and PPTX files through machine translation with translation memory and glossaries to validation and packaging of finished files. A dedicated validator workspace enables language specialists to review, edit and approve segments with full history and audit trail. The solution is implemented end-to-end with focus on scalability, traceability and correct handling of fonts – also for non-Latin languages – and is in operation while testing, monitoring and selected UI functions continue to be developed.
Medical technology / health technology (Radiometer – technical documents and materials)
Translation Management System (FastAPI backend and Next.js frontend) for multilingual documents with machine translation, translation memory, glossaries and validator workspace
The core system is implemented and works end-to-end; the system handles upload, translation, validation, packaging and download while testing, monitoring and some UI functions continue to be developed
To create a central, scalable and traceable platform for translation and validation of Radiometer's documents across many languages and file formats – with focus on quality, consistency and correct handling of fonts.
Before the solution was developed, Radiometer's translation workflow was based on an older desktop tool that made handling of multilingual documents heavy and difficult to scale. Documents in different formats (IDML to InDesign, Word, PowerPoint) had to be handled manually, and there was limited support for ensuring that technical terms and formulations were used consistently across documents and languages.
Quality assurance and validation were complex because validators lacked a consolidated overview and tools to work segment-by-segment with history and comments. At the same time, it was critical that fonts – especially for non-Latin languages – were handled correctly so the finished layout remained readable and brand-compatible.
The goal of the project was to establish a modern, server-based translation management system that automates manual steps in the translation process, ensures linguistic consistency and makes quality assurance more structured.
The system was to handle multiple file formats and many languages in a uniform way – including correct font management for non-Latin scripts – and create traceability and overview of status, jobs, versions and artefacts in one common platform.
The solution is a full translation management system built on a FastAPI backend and a Next.js frontend, supported by a Postgres database and Celery/Redis for background work.
The system combines document handling, translation pipeline, translation memory, glossaries, validator workspace and font management in one architecture designed for medical technology documentation.
The system can be described as a flow from input to output, where documents move through upload, translation, validation and packaging.
Upload and setup
The user uploads documents (IDML, DOCX, PPTX or ZIP), selects target languages and gets Document and DocumentTarget records created with metadata and desired languages.
Queue and translation
Celery jobs per target language fetch files from storage, unpack them, identify text content and combine glossary protection, translation memory and machine translation. Results are stored as segments and translations in the database.
Validation and editing
Validators work in a dedicated workspace where they can read, edit and approve segments with history and comments. When everything is approved, the document switches to validated and packaging is triggered.
Packaging and distribution
Backend generates finished documents by inserting the validated translations into the right structures and ensuring correct font embedding. Output is stored in storage and users can download files via the library page.
Reuse and further work
All translated segments become part of the translation memory and can be reused in future projects. Glossaries and TM can be extended, and logging/status endpoints enable ongoing monitoring.
The system is implemented and can handle translation workflows end-to-end from upload to download of finished files. Backend API, database, file format support (IDML, DOCX, PPTX), translation memory, glossary handling, validator workspace and font management are all implemented.
The solution is in use as a functional system while work continues on extended test coverage, improved error handling, monitoring, notifications and some UI improvements. Data and experience from operation and long-term effect are still being gathered.
The project shows that effective automation of translation in a medical technology environment requires a dedicated architecture that understands both document formats, linguistic requirements and regulatory sensitivity.
An important result is that structured validation is as important as the translation itself. The dedicated validator workspace supports language specialists working systematically and documenting their decisions.
Font and layout handling for many languages cannot be treated as an afterthought – it requires its own services and models if the end result is to be readable and brand-consistent.
Scalability and traceability go hand in hand: with status tracking, job IDs, metadata and audit logs, the system can both scale up and be monitored so that control is maintained over complex translation flows.
With the current architecture there are several realistic paths for further development of the solution:
Overall, Radiometer Translation System stands as a specialised foundation for translation and validation of technical documents across languages – with an architecture that can be extended as Radiometer's needs develop.
Next step – completely non-binding
If you want an honest assessment of where technology – AI, automation, data or software – can add value to your business, you can have a short, non-binding chat with Lars. No sales scripts – just a concrete proposal on whether there's potential, and what a first safe step could be.
