Build and operate globally distributed inference systems, including routing, capacity management, scheduling, and multi-cloud fleet scaling. Develop Rust services and Kubernetes-based infrastructure running across unreliable environments. Own production systems under significant load, diagnose failures deeply, and improve latency, reliability, and performance. The role requires strong distributed-systems experience and offers substantial responsibility in a fast-growing, in-person startup environment.

Develop and optimize low-level GPU kernel and systems code for a globally distributed inference cloud. Improve GPU throughput, reduce model costs, diagnose idle hardware, and validate performance through rigorous profiling and benchmarking. The role requires strong C or C++, systems-level knowledge, and a proven record of delivering significant performance improvements. GPU or CUDA experience is not required, but operating systems, drivers, compilers, firmware, real-time systems, or inference-engine experience is preferred.