Product Director, Localisation & AI Evals
About this role
Leading a cross-functional squad of product managers, developers, and language managers, the Product Director will drive the zero-to-one build and launch of frontier agentic translation and autonomous QA systems supporting over twenty languages. OKX seeks a product leader to tackle high-ambiguity challenges in the Web3 space, where rapidly changing crypto terminology, dense trading jargon, and complex regulatory requirements demand innovative localization solutions. Instead of managing external vendors, the incoming director will architect a human-in-the-loop agentic translation system alongside an automated evaluation product that self-verifies fixes and monitors production quality.
Operating within a major global cryptocurrency exchange environment, the role requires bridging the gap between cutting-edge LLM capabilities and practical financial applications. Successful candidates thrive on building novel products at the edge of artificial intelligence, taking full ownership of business metrics rather than mere delivery volume. Ideal practitioners possess deep hands-on experience constructing large-scale agentic systems and automated evaluation pipelines, ready to ensure that localization scales efficiently without compromising user conversion or regulatory compliance.
What you will do
- Lead a squad of product managers and developers in close collaboration with language managers to ship translation and quality assurance systems into production.
- Architect and deliver an end-to-end agentic translation system across more than twenty languages utilizing human-in-the-loop mechanisms.
- Break the cost curve by maintaining flat quality while coverage scales across languages, content volumes, and release velocities.
- Integrate the new localization infrastructure cleanly with existing company content pipelines and software stacks.
- Stand up an autonomous quality assurance agent that evaluates translation and localization performance at scale through LLM-as-a-judge and evaluation harnesses.
- Define and drive hard quality metrics such as human-acceptance rates and edit effort to ensure localization standards are systematically optimized.
What you need
- Demonstrated experience as a product management leader taking frontier agentic products from zero to one.
- Proven ability to ship novel products operating at the edge of large language model capabilities.
- Strong accountability for business outcomes and measurable impact rather than output volume.
- Deep hands-on background building large-scale, frontier large language model agentic systems from scratch.
- Familiarity with harness and orchestration layers, multi-agent workflows, and automated evaluation or LLM-as-a-judge pipelines.
- Experience building quality systems, automated evaluations, or release gates for autonomous software launches.
- International product experience serving diverse user bases across multiple regions and languages.
Nice to have
- Built an agentic system from scratch
- Shipped agentic or LLM applications in production
- Finance, payments or another highly regulated industry
Benefits
- Competitive total compensation package
- L&D programs and Education subsidy for employees' growth and development
- Various team building programs and company events
- Wellness and meal allowances
- Comprehensive healthcare schemes for employees and dependants
Skills and tech stack
Questions candidates ask
Where is this role located?
The position is based in Hong Kong, Hong Kong SAR and Singapore, Singapore across multiple global offices.
What compensation and perks are available?
The company offers a competitive total compensation package, learning and development programs, wellness and meal allowances, team building events, and comprehensive healthcare schemes.
What kind of systems will I be building?
You will own the zero-to-one build of a frontier agentic translation system supporting 20+ languages and a self-improving agentic QA product that autonomously evaluates localization quality.
Do I need a background in translation or linguistics?
The role does not require a linguistics or translation background, multilingual ability, or algorithm engineering skills.