Head of AI Engineering, Shanghai
On-siteShanghai, Shanghai, China
Job Summary
Build the team from 2–3 engineers to 10–15 across model substrate and evaluation pods, setting the hiring bar and running interview slates with Shanghai recruitment. Own the diversified model substrate by onboarding open-weight models on third-party cloud and on-prem inference, ensuring the full provider matrix remains live while reducing supply risk from export restrictions. Benchmark models per business function to guide user selection and upgrades, then construct evaluation machinery with shared graders, calibrated judges, and a certification gate that replays agent suites against candidates. Optimize routing and inference costs by closing the loop from spend visibility to routing decisions proven safe by evals, partnering with global leads on day-one buildout.
Required Qualifications
- deep production experience serving and tuning large models
- inference optimisation
- GPU capacity planning
- hosting open-weight models (GLM, DeepSeek, Qwen, or similar) on cloud and on-prem
- built or run model evaluation systems at scale
- benchmark design
- LLM-as-judge calibration
- regression gating for model upgrades
- grown an engineering team from a handful of people to 10 or more
- setting and holding the hiring bar
- reason about models as cost/capability trade-offs
- defend a routing decision with eval data rather than intuition
- fluent Mandarin and English
- comfortable operating with senior stakeholders across US, EU, and APAC time zones
- Senior enough to be the first hire and set the bar
- owns a live deliverable (the substrate buildout) from day one
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.