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 RoleWe'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 ResponsibilitiesOwn 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.
You almost certainly have:
A track record of personally scaling a high-traffic, low-latency API in production, whether at a gaming company, a video streaming company, a payments company, or a hyperscaler.
Deep k8s experience, including multi-region deployments.
Comfort with SLOs and capacity planning.
Strong ownership instinct — you've taken systems end-to-end, not just contributed to them.
Bonus points for any of:
Experience deploying streaming video or audio inference models (the dream hire).
Experience with low-latency game streaming or video streaming infra.
Experience scaling GPU fleets across providers (GCP, Coreweave, Lambda, etc.).
Experience with frontier model inference (LLMs, world models, multimodal).
Experience with on-device / edge inference (ExecuTorch, Core ML, etc.).
Our stack
GPUs: GCP today, Coreweave as we scale.
Orchestration: Kubernetes, multi-region.
Models: In-house frontier research — agents, world models, video understanding.
API surface: Client libraries in TypeScript, Python, Rust, and C
In-office, 5 days/week. NYC, Stockholm, London, Paris, or Geneva.
Competitive salary and meaningful equity
Comprehensive medical, dental, and vision coverage
401(k)
Wellhub membership for fitness and wellness
Mental health support through Spring Health and Headspace
Fertility and maternal health benefits
Paid parental leave
Generous PTO, 11 paid company holidays, and paid sick time
Daily meals and commuter benefits at our NYC HQ
Learning and development stipend
Benefits vary by country and employment type.
Skills Required
- Experience scaling high-traffic, low-latency APIs in production
- Deep experience with Kubernetes and multi-region deployments
- Strong ownership in end-to-end systems
What We Do
About Us: Create gaming memories while apart — Medal enables you to reliably capture and meaningfully share online memories with friends (..that would otherwise be lost to time).
