You will lead the Model Routing & Inference team at Cursor, owning the inference platform that powers every AI interaction in the product. This team owns the full inference path: making Cursor's AI faster, more reliable, and more cost-effective at a scale few teams in the world get to operate at. Every agent session, every tab completion, and every chat message flows through your stack.

You'll set technical direction for cluster management, inference optimization, and traffic egress, building the platform that lets the rest of the company move fast without worrying about provider complexity. You'll lead a team of strong engineers, set strong direction for the business, and make the calls that balance latency, cost, reliability, and user experience across millions of daily requests.

What you’ll do

• Building and evolving our inference gateway, a single abstraction over every provider's API semantics, so model onboarding becomes a config change.
• Building the systems that dynamically select the best model for each request based on cost, latency, and quality.
• Managing GPU cluster utilization and capacity planning across providers, optimizing for cost and performance.
• Designing routing backpressure and admission control so traffic spikes don't cascade into providers.
• Hiring and growing the team: sourcing, interviewing, and closing top inference and systems talent, while developing your engineers through coaching, mentorship, and high-leverage project assignments.

You may be a fit if

• You have led engineering teams building high-throughput, low-latency distributed systems, especially in inference serving, traffic routing, or real-time data pipelines.
• You're comfortable reasoning about cost/performance tradeoffs at scale (GPU utilization, provider economics, capacity planning) and making decisions with incomplete information.
• You have strong software engineering fundamentals and enjoy shipping production systems that handle millions of requests.
• Experience with model serving frameworks (vLLM, TensorRT-LLM, TGI), load balancing, or building resilient multi-provider architectures is a plus.
• You make good calls in the gray area: weighing reliability, cost, latency, and user experience when there isn't a single 'right' answer.

XML job scraping automation by YubHub