yoinka

Embedded Linux Software Engineer, XFRA

Span

San Francisco, CAFull TimeMid
Sign in to applyVerified
Location
San Francisco, CA
Employment
Full Time
Work model
On-Site
Level
Mid
Posted

Skills

LinuxPythonRust

About this role

Our Mission

SPAN is enabling electrification for all ⚡ XFRA by SPAN is the first distributed data center using existing infrastructure and underutilized power capacity to meet expanding AI compute demand. The Product XFRA is a new business inside SPAN. We are leveraging SPAN's smart panel and digital power controls to enable distributed AI inference compute infrastructure: GPU compute nodes deployed into residential, commercial, and industrial sites, orchestrated as a single fleet. It is a genuine v0 product. We now need to build a team to stand up a manufacturing and deployment process for XFRA, develop a production-grade node management and orchestration layer, and continue growing revenue by selling compute and AI services. For more details, check out our website and whitepaper: https://www.xfra.ai/ The Role XFRA is a small, fast-moving team building a new product from nothing. We are looking for a Linux Software Engineer to build and own all software that runs on an XFRA node: the software image on our GPU servers, the node controller that connects those servers to the thermal and electrical systems around them, and the node-side interfaces that our orchestration layer and our customers use to run AI workloads on the hardware. Over time, your software will manage hundreds of thousands of nodes, representing many gigawatts of cutting-edge distributed AI compute capacity.

Responsibilities

Own the XFRA node software image end to end: kernel, drivers, package and repository pinning, system services, and a reproducible, version-controlled build Release node images on a regular cadence through our cloud deployment path, designed so a failed update degrades gracefully rather than bricking a node hours from the nearest engineer Build and own the node controller: sensor and actuator interfaces, local control loops, and its telemetry contract with the cloud Integrate nodes with SPAN panels so workload can be curtailed, shifted, or offloaded in response to energy events and site power limits Build the node-side API that the orchestration layer talks to, and the interface customers use to schedule and run AI workloads on the hardware Document what you build: architecture, interfaces, and operational procedures About You Required Qualifications 4+ years building and shipping production systems software on Linux Deep Linux internals: kernel configuration and drivers, boot and init, systemd, filesystems, networking Experience building and releasing a Linux image or firmware for fielded hardware, including a remote update mechanism that is safe to run unattended Strong grasp of at least one of C, Rust, or C++, as well as Python Hands-on hardware integration: sensors, actuators, buses, and debugging across the software/hardware boundary with a scope, logic analyzer, and a datasheet Real operational experience owning remote systems nobody can physically touch, including participating in the team on-call rotation Bonus Qualifications Prior experience as a founding or early engineer, or taking a 0-to-1 product from nothing to production You use AI tooling heavily and deliberately to increase your own output OS build systems and the CI around them: Yocto, Buildroot, Debian/Ubuntu packaging A/B update systems such as OSTree, Mender, RAUC, or SWUpdate Server platform bring-up: BMC and out-of-band management, UEFI/BIOS configuration, PXE or netboot provisioning Thermal or power control systems, control-loop design, or liquid cooling Energy, electrical, or power-electronics systems; industrial protocols such as Modbus, CAN, or RS485 GPU or AI inference infrastructure: NVIDIA driver and CUDA bring-up, multi-GPU systems, GPU partitioning, GPU-aware container runtimes Host and platform security: intrusion detection, integrity measurement, hardening, secrets handling on hardware you do not physically control Telemetry pipelines and pub/sub messaging (MQTT or similar) from remote devices Life at SPAN Headquartered in San Francisco’s vibrant

Listing verified . Applications go through the company's official careers site.

← Back to Yoinka