About DeepReference
DeepReference is a place to translate documents, review the result, and keep terminology consistent from one job to the next.
Team
DeepReference is built by a small team covering translation, language operations, data science, business, and platform engineering.
Project Lead
Guides product direction, platform delivery, language coverage, and document workflow decisions across DeepReference.
Language Technology & QA Engineer
Works on translation data workflows, quality checks, prompt behavior, terminology handling, software testing, and model-output review.
Operations & Customer Success
Helps shape customer workflows, onboarding, business development, operational planning, support priorities, and practical use cases for translators and teams.
What we are building
DeepReference is being built for document translation that will be checked before it is delivered.
A document job includes more than target text. Terminology, formatting, review comments, earlier approvals, and revision history all affect the finished file, so the workspace keeps those parts with the job.
The model prepares a translation; a person decides whether it is ready. Our work is to give that reviewer useful context and reliable controls without making the document harder to handle.
The translation engine runs with the settings and language assets you choose. You can apply a glossary, select a domain and tone, and use Natural or Literal mode according to the job.
After translation, you can work segment by segment, check a translated DOCX in its layout, compare document revisions, and save approved sentences to Translation Memory.
DeepReference currently supports 45 languages and 12 subject domains. The domain settings cover legal, medical, financial, technical, government, manufacturing, marketing, automotive, energy, telecom, insurance, and pharmaceutical material.
45 Languages
Support for major European, Asian, Middle Eastern, and African languages.
Document Workflows
Translate office files, review their segments, and download the completed document from the same job.
12 Subject Domains
Give the engine subject context for legal, medical, technical, and other specialist material.
EU Data Residency
EU-hosted processing, account controls, and clear deletion options.
Data Security
Sensitive documents need careful handling. DeepReference uses encrypted transport, browser-side content sealing for supported API traffic, account-based access, and controlled processing for uploaded files.
Customer documents are processed to provide translation, editing, terminology, and review workflows. They are not used for model improvement unless the account owner explicitly opts in.
You stay in control of your workspace data. Documents, generated translations, glossary entries, and translation memory data can be removed from your account when you no longer need them.
Our Technology
DeepReference uses a dedicated translation engine connected to workflow tools for documents, glossary, translation memory, editing, and revision comparison.
It is designed for work that needs more than a first-pass translation: terminology decisions, review, corrections, and reuse all stay attached to the job.
Trust and Data Protection
We publish clear information about security, privacy, data handling, and quality controls as the platform develops.