Hennessy Ng

Product owner and builder for an AI-native localization management platform — owning discovery, roadmap and delivery while writing production code across the stack.

• Own the roadmap and product definition for an AI-native localization management platform, translating localization operations pain points into shipped capability.
• Design and build end-to-end Django / Vue.js workflow pipelines, delivered through GitHub Actions CI/CD.
• Design RAG systems and AI agents that automate the translate–edit–proofread (TEP) workflow, keeping terminology and style consistent across locales.
• Develop and optimise localization prompts, and build LQA tooling to score machine-translated output for accuracy.
• Evaluate cloud and open-source LLM options against cost, latency and quality to keep AI localization workflows efficient.
• Architect cloud infrastructure on AWS, running microservices orchestrated with Kubernetes.

Built the NLP, OCR and machine translation automation that moved a traditional language services business onto cloud-native infrastructure.

• Designed and shipped an OCR + machine translation pipeline processing 500+ scanned PDFs (~5,000 pages) daily from Japanese to English.
• Built bilingual term extraction using BERT embeddings and semantic similarity, generating termbases that fed both MT training and linguist consistency.
• Developed multilingual web crawling and language detection pipelines to acquire and clean parallel corpora for MT training.
• Migrated desktop-bound language tooling to cloud-based RESTful APIs, integrating with CMS platforms and cloud MT vendors.
• Trained custom MT engines on validated post-edited output, recursively improving quality with roughly 40% of production data validated and reusable.

Owned translation quality assurance, and introduced neural machine translation and OCR into the QA workflow.

• Led the translation QA function, defining the standards and review process applied across delivery.
• Introduced neural machine translation and OCR into production workflows, shifting the team toward post-editing rather than translation from scratch.
• Ran linguist recruitment and training, building post-editing capability across the team.

Computer-aided translation tooling, terminology management and hands-on linguistic delivery.

• Administered and supported computer-aided translation tooling across the delivery team.
• Owned terminology management — termbase creation, curation and maintenance for corporate accounts.
• Delivered translation and proofreading work directly, grounding later automation in real linguist workflow.