Multilingual translation and localization of IVD documentation package (74 languages) with terminology, CAT and QA

Initial situation
The company produces an IVD product - an analyzer and a test kit for clinical diagnostics. Now all documentation is supported in EN and partially in ES/DE in the form of sources in Markdown and interface strings in JSON. It is required to prepare a complete set of translations into 74 languages ​​for release in several regions, preserving terminology and legally significant wording.

Subject area and technical environment
- IVD medical devices, clinical diagnostics, labeling, user safety, warnings
- Sources: Markdown (IFU/Quick Start/Labeling notes documents), JSON (UI strings), CSV (glossary), PDF references of older versions
- Tools: CAT with TM/terminology and QA support (e.g. Trados Studio, memoQ, Phrase TMS - you can offer your own stack), Xbench/Verifika or equivalent for automatic checks, Git for versioning

What needs to be done in stages
1) Pre-translation analysis
- Assess the volume and repetition, collect TM from the old translations provided (if quality allows) and configure the project for 74 languages
- Create/clarify a glossary: ​​key IVD terms, names of tests, units of measurement, warnings and contraindications

2) Translation and localization
- Translate Markdown documents and JSON interface strings into 74 languages ​​using TM and glossary
- Save Markdown structure, links, code blocks, placeholders in JSON (for example {0}, %s), variables and escape sequences
- Ensure consistency of terms between documentation and UI

3) Linguistic and formal verification
- Run a QA report for each language: terminology inconsistencies, omissions, inconsistent numbers/units, damaged placeholders, length of UI lines (where critical)
- Conduct selective proofreading of high-risk sections: warnings, precautions, restrictions, interpretation of results

4) Delivery and support of imports
- Serve translations in original Markdown and JSON formats without changing the file layout
- Attach TMX and glossary (TBX/CSV) in an updated form
- Attach QA reports and a list of controversial areas with questions/solutions

Specific result
- A package of localized files for 74 languages, ready to build documentation and import strings into the product
- Updated TM and glossary suitable for future releases

Restrictions
- You cannot change JSON keys, Markdown structure, file names and their encoding (UTF-8)
- Approved warnings must not be rephrased in a way that changes the meaning or level of severity
- You cannot use machine translation without post-editing to a professional level and without passing QA

Measurable acceptance criteria
- 0 damaged placeholders and variables in JSON, 0 changed keys
- 0 broken links and 0 Markdown violations, rendered correctly in our pipeline
- QA: no critical errors (missing translation, placeholder mismatch, number mismatch, terminology forbidden term). Up to 5 minor comments per language are allowed, corrected within 48 hours
- Terminology from the agreed glossary is observed at least 98% according to a random check on 500 segments

Input data from the customer
- Git repository with sources, old PDFs and partial translations, current glossary in EN, list of target 74 languages ​​and locales, UI screenshots for context

Expected Communication
- Questions on terms and controversial wording are collected in batches once a day, so as not to slow down the translation flow