Head of Inference

Canyon Code is a Silicon Valley startup building AI infrastructure and an inference platform for multi-agentic applications. The Head of Inference will own the end-to-end inference function, including GPU capacity procurement, open-weight model deployment and optimization, reliable endpoint serving, cost efficiency, and building the inference team.

Responsibilities:

• Capacity. Procuring bare-metal GPU from neoclouds and colocation facilities: evaluating providers, negotiating terms, taking delivery, and getting it production-ready
• Models. Bringing open-weight models up on that capacity and keeping them optimized as the mix changes
• Endpoints. Serving them reliably, at the cost per token that makes our economics work
• The team. Hiring and building the inference team underneath you as the motion scales