We make the best chips physically possible for the large model needs of frontier labs.
About the role
Architect, design, and implement the Linux kernel driver stack for MatX's AI accelerators (PCIe endpoint devices), including memory management, DMA, interrupt handling, and command submission paths Build the low-level userspace libraries that sit directly on top of the driver — ioctl/sysfs interfaces, memory mapping, direct chip access utilities
What they're looking for
- BS or higher in Computer Science, Electrical Engineering, or equivalent practical experience, with 8+ years in low-level systems software
- Strong hands-on Linux kernel development experience — you've written and shipped device drivers, ideally for PCIe devices
- Strong C programming skills, comfortable in userspace and kernel space
- Able to read hardware datasheets and register definitions and program a device from them, without waiting for someone to explain the hardware
- Deep understanding of computer architecture and OS internals: MMIO, DMA, interrupts, IOMMU, memory hierarchy, processes/threads, virtual memory
- Experienced debugging complex hardware/software interaction issues with limited visibility
More about this role
MatX's mission is to make the world’s best AI models run as efficiently as allowed by physics, bringing the world years ahead in AI quality and availability. MatX is seeking System Software Engineer to join our team as we create best-in-class silicon for high-performance and sustainable GenAI. Successful candidates for these roles will be responsible for delivering performant and functionally accurate silicon for MatX products across compute, memory management. High-speed connectivity and other key technologies.
The MatX host system software team owns everything that makes our AI silicon and systems usable: from Linux kernel drivers up through node and cluster management. The team also co-owns the BMC/OpenBMC firmware stack, with dedicated firmware engineers, so host software and out-of-band management are designed together rather than bolted together. We're looking for self-driven engineers who can take a hardware spec and a register map and just start building — prototype drivers, low-level utilities that talk directly to the chip, daemons, and tooling — with minimal hand-holding. Each engineer on this team has a primary focus area, but ownership of overlapping components is...
Browse similar: Startup jobs · San Francisco Bay Area