TermAuditor: AI-Powered Terminology Governance at Global Scale
Member Webinar
Uber operates in 60+ languages across multiple lines of business, where small terminology errors can erode brand consistency and create regulatory exposure — at a volume no manual review process can keep up with.
This session presents TermAuditor, an agentic AI system built by Uber's Localization Technology team that automatically validates translated content against an approved glossary and corrects it in context, in real time.
We'll walk through the two-stage detection-and-correction pipeline, the data-driven engineering decisions that cut processing cost by ~76% and eliminated a key reliability risk, and where the system is headed next with a more structured, persona-aware terminology schema. Attendees will leave with a practical blueprint for scaling terminology QA with AI while keeping human oversight where it matters most.
Three takeaways
1. AI can enforce terminology consistency at a scale manual review can't reach — a two-stage detection-and-correction pipeline validates translations against tens of thousands of approved terms in seconds, respecting context and persona rather than applying one-size-fits-all rules.
2. Efficiency gains come from data, not guesswork — production usage data drove concrete engineering optimizations (a ~76% reduction in processing cost per request, a ~5x cut in redundant review, and elimination of a silent reliability failure mode), all validated before rollout.
3. The next frontier is structured, persona-aware terminology data — moving from legacy prose-based glossaries to a machine-readable schema reduces semantic collisions and hallucinated corrections, and points to where terminology management is heading industry-wide.
Target audience
- Localization program managers
- Terminology and glossary managers
- Localization engineers
- Language quality leads
- Companies scaling translation across many languages and lines of business — particularly those evaluating how AI/LLMs can be applied responsibly to terminology QA and glossary governance.
Registration Options
|
Registration Options
|
Price |
|---|---|
|
MEMBER TICKET
|
FREE |
|
NONMEMBER TICKET
|
$35.00 |
James Lin, Senior Manager of Localization Technical Program Manager at Uber, brings over 20 years of expertise in internationalization and localization, with a strong focus on language and technology adoption/expansion, HRL and LRL language support, and a technology-centric approach.
He has spearheaded multiple initiatives to optimize Uber’s localization pipelines, achieving significant cost savings of millions of dollars through an in-house orchestration layer, custom quality thresholds, and confidence scores in MT aggregators. His work emphasizes adaptive, smart translation routing via data-centric strategies and technology integration. Moreover, James has driven the adoption of MT/LLM domain models across 100+ languages, effectively integrating advanced machine learning techniques with internationalization best practices to enhance both scalability and impact.
By blending cutting-edge ML approaches with comprehensive i18n methodologies, James’s innovative strategies ensure that Uber’s localization efforts remain efficient, scalable, and transformative.
A certificate of attendance is available on request for those who join the live session.
For More Information:
