Development of an Advanced Nlp/Llm + Rag Localization Platform

We are seeking experienced developers to build a business-aware localization platform leveraging nlp, llm, and rag technologies. This platform will produce tenant-specific translations using approved glossaries, proprietary corpus data, translation memory (TM), domain rules, and configurable prompts, moving beyond a generic translation chatbot.

The primary goal is to enhance localization quality, consistency, scalability, and auditability, while simultaneously reducing costs and turnaround times.

Key functionalities and workflow requirements include:
- Tenant identity validation, authentication, subscriptions, quotas, and permissions management.
- Loading of customer-specific glossaries, domain rules, and prompts.
- Retrieval of relevant corpus and translation memory records.
- Generation of localized content.
- Validation of terminology and formatting.
- Confidence score calculation for generated content.
- Triggering human review for low-confidence or sensitive outputs.
- Recording of usage and audit data.
- Returning localized output along with comprehensive metadata.

The platform must enforce strict tenant isolation to ensure that customer data, glossaries, prompts, and translation memories remain entirely separate. A glossary-first approach is crucial, ensuring that approved terms are utilized, forbidden terms are blocked, and any conflicts are promptly flagged. The RAG component will retrieve domain-specific examples prior to content generation, thereby improving terminology consistency and providing traceable references for all translations.

Quality scoring will evaluate glossary compliance, retrieval quality, and overall domain fit. Outputs with low confidence or those identified as sensitive will be routed for human review. Approved edits from human reviewers will be used to update the translation memory for future reuse. The system should support offline SQLite bundles containing authorized tenant data, track usage for accurate billing and reporting, return business metadata including confidence and traceability, handle business-specific errors gracefully, and meet all enterprise requirements for secure, auditable, and cost-effective localization solutions.