About the role
About General Intuition We are the frontier research lab dedicated to building foundation models for environments that require deep spatial and temporal reasoning. For the past year, we've been pushing the forefront of AI across agents capable of navigating space and time, world models that provide training environments for those agents, and video understanding models with a focus on transfer to the real world. We raised a seed round of $133M from General Catalyst and Khosla to discover the next generation of intelligence.
The Role We're hiring an Infra Engineer to own General Intuition's API. Our research team builds frontier models — agents that reason about space and time, world models, video understanding. Your job is to turn those models into a production API that developers love: low-latency, highly available, billing-grade reliable, and able to scale from our first hundred users to tens of thousands of concurrent ones. You'll work directly with the founding team. You'll own the API end to end: the client libraries developers integrate with, how we receive frames from clients and stream actions back, how requests route to the right GPU, how sessions spin up and tear down, how k8s clusters get stood up in new regions, and how our GPU fleet scales. This is a true generalist infrastructure role. We are not looking for a pure API person or a pure GPU person — we are looking for someone who is exceptional at both, and who wants to own the entire surface end-to-end. Key
Responsibilities Own the video streaming protocol . Orchestrating how we receive frames from clients and route them to servers as efficiently as possible. Own the runtime layer of our API. Stateful request routing, GPU session lifecycle, inference orchestration — the whole runtime stack. Scale our k8s footprint across regions. Lead new regional deployments. Own the GPU hosting strategy. Move us from dozens of GPUs today to potentially thousands (and beyond) without breaking the bank or the latency budget. Drive latency and throughput. Own the inference-performance backlog Partner with product engineering. Work closely on developer-facing reliability, observability, metering, and billing-grade uptime.
