Distributed Systems Engineer

Relace•San Francisco, CA

1d•Onsite

About The Position

Relace is building a globally distributed inference cloud that runs customer requests on its own GPU fleet across multiple clouds. The company operates Kubernetes across its fleet and builds its own stack, with performance, price, speed, and reliability as core priorities. It processes more than 800 billion tokens per day, has grown threefold in the last three months, and is a top provider by volume on OpenRouter. Relace raised a $23 million Series A led by Andreessen Horowitz, with participation from Matrix Partners and Y Combinator. The Distributed Systems Engineer will work on the systems that run globally distributed inference, including the routing layer that selects machines for requests, handling demand that exceeds available capacity, and scaling the fleet across clouds as demand changes. The work involves Relace’s Rust service and Kubernetes services that manage a multi-cloud fleet. This is a senior role based in San Francisco, in person five days per week, with competitive compensation plus equity.

Requirements

Around five or more years building production systems where latency and failure behavior mattered, such as serving systems, control planes, schedulers, databases, networking, or trading systems.
Strong knowledge of a systems programming language such as Rust, C++, or Go.
Ownership of a production system that ran under load.
Ability to discuss in detail a production failure that the candidate personally debugged.
Kubernetes experience at the level of having extended it, particularly on low-reliability infrastructure.
Distributed systems instincts.

Nice To Haves

Experience with a scheduler or load balancer as a substantial project rather than a side task.
Experience working with spot or preemptible capacity.
Exposure to AI serving systems; this is explicitly optional.

Responsibilities

Develop and operate the routing layer that determines which machine handles each request.
Build systems for handling demand that exceeds available capacity.
Build systems that scale the fleet across clouds as demand changes.
Work on Relace’s Rust service and Kubernetes services managing a multi-cloud fleet.