Founding Infrastructure Engineer

Role overview Own the infrastructure that serves open AI models, from GPU capacity through the production inference stack. You will build and operate systems as they move from private beta to significant live traffic, balancing speed, cost, and reliability in a hands-on founding engineering role.

Responsibilities - Design, build, and operate GPU infrastructure for high-throughput, low-latency inference. - Run and tune serving systems such as vLLM, SGLang, TensorRT-LLM, or custom engines. - Lead reliability work across monitoring, capacity planning, incident response, and on-call practices. - Work with researchers to bring hardware and software experiments into production. - Coordinate GPU capacity with external partners and manage secure, performant networking across cloud providers. - Improve routing, cache use, scheduling, observability, and deployment processes as demand grows.

Requirements - 10+ years of hands-on SRE, platform, backend infrastructure, or related production systems experience. - Experience building in an early-stage startup and operating systems at substantial scale. - Track record with high-traffic, highly available services, ideally with availability targets around 99.99%. - Recent production experience in GPU or AI infrastructure; inference serving is especially relevant. - Strong knowledge of Kubernetes, Linux, networking, and distributed systems. - Comfortable writing and operating code directly, carrying an on-call pager, and communicating clearly.

Nice to have - GPU cluster, HPC, or machine learning platform experience; familiarity with NCCL, RDMA/InfiniBand, CUDA, or bare-metal systems. - Experience with LoRA-aware routing, KV cache locality, inference engine optimization, or production rollout automation.

Benefits and work setup - Remote role for candidates in UTC−7 or UTC−8, aligned to West Coast hours. On-call is shared across US and European time zones.