Software Engineer, Embedded BSP/Kernel (Robotics OS)

About the role

Meta Robotics Studio builds consumer humanoid robots that bring Meta's AI into the physical world. Our robots run on NVIDIA Jetson Thor with Meta-designed carrier boards and JetPack OS. The Robotics OS team owns everything between the silicon and the robot learning stack. On one chip, our robots run real-time control loops over EtherCAT while large vision-language-action (VLA) models run on the GPU. Our job is to make that computer do both, deterministically, every time it boots. We are hiring a software engineer to help architect and evolve the robotics OS, Meta's build of NVIDIA Jetson Linux for Thor-class robot compute. Working with the team, you will help shape the multi-year technical direction for the robot's operating system, including how we build, release and update it, how it shares one computer between real-time control and AI workloads, and how it scales from a lab of prototypes to a large fleet of production robots. You will work closely with hardware, firmware, controls, and AI engineers across Meta, and with NVIDIA's Jetson engineering teams as one of several technical points of contact on the platform. This is a hands-on role: when a new board spin won't boot or the control loop misses a deadline, you will be in the kernel log alongside the rest of the team.

Responsibilities

Contribute to the technical strategy and multi-year roadmap for the Robotics OS. This covers the board support package (BSP), kernel, OS image, update system, real-time behavior, GPU platform, and fleet tooling across every robot program and compute platform (AGX Thor, IGX Thor, and future NVIDIA silicon) Architect the next-generation OS image, including the move to a Yocto / meta-tegra based build. Make it reproducible, hermetic and secure, and make it fast for many engineers to iterate on Define and drive the production update strategy: atomic A/B and over-the-air updates, rollback, secure and verified boot, fusing and signing, and factory provisioning, so we can ship to a growing fleet safely Shape the platform's real-time architecture. Help define how CPU cores, interrupts, memory and the GPU are divided between control loops at kilohertz rates and large AI models. Set the latency budgets and the measurement infrastructure that proves we meet them Collaborate with NVIDIA's Jetson engineering teams alongside other Meta engineers: share requirements, push fixes upstream, and help plan how we adopt new Jetson Linux / JetPack releases with minimal fork debt Partner with electrical, firmware, and mechanical engineers on new compute and carrier-board designs. Weigh in early on SoC configuration, interfaces, power and thermal so boards are easier to bring up and software stays simpler Drive bring-up of new boards and modules end to end: boot chain (MB1/MB2, UEFI), BPMP and device tree, pinmux, PCIe/USB/Ethernet, power profiles, and the drivers for robot hardware (EtherCAT, USB-C power delivery, Wi-Fi/Bluetooth, sensors, cameras) Help set the quality bar for the RT kernel and out-of-tree drivers: kernel configuration, ABI compatibility, and a plan to upstream changes or align with Meta's kernel instead of carrying long-lived patches Build the reliability and observability foundation: crash and kernel-log capture, symbolization, boot-time tracing, health telemetry, and fleet triage tooling, so that any robot failure can be diagnosed remotely Raise the bar for the engineers around you through design reviews, mentorship, and technical hiring. Grow OS and embedded talent on the team, and help shape on-call and operational practices