Early Career Device Software Engineer

Early Career Device Software Engineer

Location:

London or Bristol - Hybrid
About Fractile

Fractile was founded in 2022 on the bet that, eventually, the world’s most capable AI systems would be limited in their impact by the time taken to produce useful outputs. We bet everything on the logical conclusion: that the only way to truly unlock this latent value, to make speed viable at scale, was to radically re-invent the hardware that we run our frontier AI models on.

Ever since, we have been building chips and systems that tackle this problem: how to efficiently generate output at thousands of tokens per second, while handling the complexity and capacity challenges of operating large models at very long contexts.

The workloads that push to the limits of the current frontier are already transformational; it is the technical and economic limits on inference speed that are constraining progress. The defining work of the 21st century will be marked by the engine of inference delivering immense and diffuse chains of intellectual inquiry, in drug discovery, in software engineering, in materials discovery, in any field where progress is driven by deep reasoning and intelligence to resolve complex problems.

Fractile is seeking to increase the clock speed of global progress, one chip at a time. We’ve recently raised $220M from investors including Founders Fund and Accel and our most important work lies ahead. Join us!

In this role you’d join small cross‑functional teams alongside engineers who have shipped kernel drivers, optimised code on real accelerators, built simulators, or brought up silicon before. They’re excellent at what they do and want you to be too.

Depending on where your interests and strengths are, you might:

• Write and debug Linux kernel drivers (in Rust!) for PCIe, DMA and memory subsystems on novel SoC platforms
• Optimise compute kernels at instruction level on RISC‑V and our own ISA where you write software that runs at the hardware’s limits
• Build firmware for bare‑metal and RTOS, as well as embedded Linux systems for the systems that keeps our hardware running safe, secure and reliably
• Build and extend our QEMU‑based functional simulator to enable software development before the silicon arrives
• Bring up software and hardware together, from first simulation runs to the first real boards
• Debug across the hardware and software boundary: kernel crashes, race conditions, memory errors, device bring‑up

• You think below the abstraction layers. When something is slow or broken, you want to know why, at register and instruction level, and you don’t stop until it works and you’ve measured it.
• You’ve gone genuinely deep somewhere. You’d likely be strong in C, C++, or Rust, have worked seriously in Linux, bare‑metal, or an RTOS, and know at least one architecture well enough to reason about what the hardware is doing.
• You’re collaborative: you share what you know and are just as ready to ask about what you don’t.

The people who stand out tend to have done more than their coursework required: a Linux driver written for some obscure bit of hardware just to watch it enumerate, an OS built from scratch that boots on a Raspberry Pi and switches between two threads, a compiler that lowers a small language down to RISC‑V just to see your own code become instructions, or a hot loop rewritten in assembly for the satisfaction of watching the profile drop.

It doesn’t need to be large or polished. What it tells us is that you’re genuinely engaged with how things work underneath.

A degree in Electronic Engineering, Computer Engineering, or Computer Science is common here, but, you may have built the same depth through industry experience or open source work.

This is a young company doing hard things. We started in 2022 and we’re now 110+ people, balanced across hardware and software. We’re building the software layer for a chip as the chip takes shape, so you’ll often be writing software ahead of the hardware: against a specification first, then a simulator, then early boards. You’ll work with incomplete information and watch it get sharper with every iteration.

That…