GPU Kernel Engineer

Description

About Relace

Relace is building a globally distributed inference cloud. Customers send us requests, we run them on our own GPU fleet around the world, and we compete on price, speed and reliability against companies far larger than us. We have customers and revenue today, we’re venture backed, and the team is still small, so you’ll own whole systems, not slices of them.

We build our own stack, and the fleet runs on Kubernetes across multiple clouds. Performance is the business: ten percent faster is ten percent more revenue. If you like making things fast, this is about as direct as it gets. You can trace a change you shipped this week to what it earned.

Some numbers: we process 800B+ tokens a day, we’ve grown 3x in the last three months, we’re one of the top providers by volume on OpenRouter, and we raised a $23M Series A led by Andreessen Horowitz, with participation from Matrix Partners and Y Combinator.

Responsibilities

• You’d work at the bottom of the stack, writing and tuning the code that runs directly on the GPU: getting more work per second and per dollar out of the hardware, shrinking models so they’re cheaper to run without making them worse, and hunting down the many boring reasons a GPU sits idle. We don’t require GPU or CUDA experience. We’d rather have an excellent low-level systems engineer with no GPU background than an average CUDA programmer.
• What we need to see
• Five or more years of low-level performance work where the speed of the system mattered: operating systems, drivers, databases, compilers, trading systems, inference engines.
• Strong C or C++ and real comfort at a low level: memory, concurrency, caches, hardware behavior.
• A track record of making something significantly faster, where you can explain exactly why it was slow and what you changed. This matters more than anything else on this list.
• The habit of measuring. You profile before and after, and you distrust numbers from a sloppy test setup.

Requirements

• Operating system, hypervisor, driver, firmware or real-time work; performance tools you built for other engineers; compiler work; actual GPU experience, in industry or open source. We can teach you the GPU. We can’t teach performance instincts.
• Working here
• San Francisco, in person, five days a week, because the hard problems get solved at a whiteboard. This is a senior role, with competitive compensation plus equity.