Embedded Systems Engineer, Humanoid Robotics

FieldAI•Boston, MA

22d•Onsite

About The Position

Field AI is transforming how robots interact with the world by building risk-aware, reliable, and field-ready AI systems. This role involves developing computing systems for humanoid robots, encompassing compute platform design, firmware and BSP bring-up, and kernel-level Linux work. The focus is on a robot payload that processes sensor data (LiDAR, camera, IMU, tactile) to output joint manipulation and locomotion commands. The engineer will collaborate with the ML team to ensure the compute platform meets their performance needs and will work to generalize solutions across different humanoid robot platforms. Close collaboration with mechanical, electrical, and ML teams is expected for integrated, safety-conscious solutions.

Requirements

B.S., M.S., or Ph.D. in Computer Engineering, Electrical Engineering, Robotics, or a related field.
Minimum of 3+ years of hands-on experience with embedded systems.
Proficient in C++ and Python for embedded and application-level development.
Experience bringing up and customizing bare-metal and RTOS firmware, Linux kernel and device drivers, and board support packages (BSPs), across platforms such as Jetson or custom SBCs.
Experience developing drivers for cameras, LiDAR, IMUs, tactile sensors, or motors.
Familiarity with GPU, TPU, or NPU offload and frameworks such as CUDA or TensorRT for edge inference.
Hands-on experience with Ethernet, CAN/CAN-FD, SPI, I2C, UART, USB, or PCIe, wireless links, and timing protocols such as PTP.
Familiarity with ROS/ROS2 and DDS for exposing sensor and actuator interfaces.
Experience designing compute hardware and firmware under tight size, weight, and power (SWaP) constraints.

Nice To Haves

Experience developing embedded systems, firmware, or drivers for humanoid or other legged robot platforms.
Familiarity with joint manipulation, motor control, and how sensor data flows into robot commands.
Experience with Linux kernel modules, device driver development, kernel-level debugging, PREEMPT_RT, and deterministic, low-jitter timing.
Experience implementing e-stop circuitry, safety monitoring, or other safety-critical embedded systems.
Experience with functional safety standards such as ISO 13849 or ISO 10218.
Experience working directly with ML or perception teams to meet model latency, memory, and throughput requirements.

Responsibilities

Design and select embedded compute platforms (ARM, SoC, microcontrollers) balancing capabilities against SWaP constraints.
Write and customize bare-metal and RTOS firmware, board support packages (BSPs), and bootloaders.
Work at the Linux kernel level to support real-time performance and reliable operation.
Conduct thermal profiling, power draw analysis, and latency measurement, and implement watchdogs and health checks.
Adapt, integrate, and develop drivers for cameras, LiDAR, IMUs, motors, joint actuators, and tactile sensors.
Bring up wired (Ethernet, CAN, GMSL, SPI, I2C) and wireless interfaces with deterministic timing (PTP, PPS).
Partner with the ML team to ensure the compute platform meets their latency, memory, and throughput needs.
Build the data pipeline connecting sensor input to joint manipulation commands.
Support accelerated inference on the backpack for real-time ML model execution.
Expose driver and sensor data through ROS/ROS2 and DDS interfaces.
Design the backpack's compute and software architecture to generalize across different humanoid robot platforms.
Work closely with mechanical, electrical, and sensor engineers to develop a tightly integrated backpack payload.
Lead the technical direction of backpack compute development.
Implement e-stop circuitry and safety monitoring on the backpack platform.