Senior Software Engineer, Robotics & Physical AI Integration

Goddard•Wilmington, MA

23h•Onsite

About The Position

We are looking for a Senior Software Engineer to own the integration layer between our AI/ML models and the physical systems they control. This is not a role for someone who writes glue code and calls it done. You will be expected to own the full stack from industrial fieldbus communication and real-time control loops up through the software interfaces that put trained models to work in the real world. You will work directly with our Machine Learning Engineer to define inference contracts and latency budgets, and with our hardware and mechanical teams to understand what the physical system can and cannot tolerate. If you have strong opinions about how software should behave when it controls hardware that moves, and you debug problems that span firmware, operating systems, and silicon without losing patience, you will thrive here.

Requirements

5+ years in systems or robotics software engineering with a demonstrated track record of shipping software that controls physical hardware in production.
Strong proficiency in C and C++; Python for tooling, scripting, and test automation.
ROS / ROS 2: Hands-on experience writing nodes, services, actions, and hardware interface layers; comfortable with launch systems, parameter management, and tf2.
Industrial Robotics: Hands-on experience interfacing with industrial robot controllers from one or more major vendors (FANUC, KUKA, ABB, Universal Robots, Yaskawa/Motoman, or equivalent); working knowledge of robot coordinate frames, kinematics, end-of-arm tooling interfaces, teach pendant workflows, and vendor-specific programming environments (e.g., Karel, KRL, RAPID, URScript).
Industrial Protocols: Practical experience implementing at least one industrial fieldbus protocol — EtherCAT, CANopen, Modbus, or Profinet — in a production system.
Embedded Interfaces: Working knowledge of SPI, I2C, UART, CAN, and GPIO, and experience debugging communication failures at the signal level.
Real-Time Systems: Experience with real-time Linux (PREEMPT_RT or Xenomai) or an RTOS (FreeRTOS, Zephyr, or QNX) for deterministic control, with an understanding of how to measure and bound latency.
Functional Safety: Familiarity with functional safety standards relevant to machinery or software in safety-critical systems (e.g., IEC 62304, ISO 13849, or IEC 61508), with an ability to translate safety requirements into software constraints.
Debugging: Demonstrated ability to isolate failures in systems where the root cause may be in software, firmware, hardware, or the interface between them.
Software Engineering: Solid fundamentals — Git, code review, unit and integration testing, CI/CD — applied to systems code, not just application code.

Nice To Haves

Experience integrating machine vision systems using GigE Vision, USB3 Vision, or similar standards; familiarity with OpenCV or other vision processing libraries.
Working knowledge of motion control concepts: servo tuning, trajectory generation, path planning, and force/torque sensing and control.
Experience consuming ML inference runtimes (TFLite, ONNX, TensorRT) within a control or perception pipeline, and an understanding of the operational constraints they impose.
Exposure to simulation and digital twin environments such as Gazebo, NVIDIA Isaac Sim, or MoveIt for offline testing and development.
Experience in a startup or small-team environment where you own architecture decisions and build tooling and process alongside the product.

Responsibilities

Design and implement software that integrates ML model inference outputs with physical hardware, including motion controllers, servo drives, actuators, and industrial I/O.
Develop and maintain ROS 2 nodes, hardware abstraction layers, and device drivers for robotics and automation hardware.
Implement and own communication stacks across industrial fieldbuses including EtherCAT, CANopen, Modbus, and Profinet.
Interface with industrial robot controllers (e.g., FANUC, KUKA, ABB, Universal Robots, Yaskawa) via vendor SDKs, proprietary communication interfaces, and standard industrial protocols; translate controller capabilities and constraints into software integration requirements.
Interface with low-level embedded hardware over SPI, I2C, UART, GPIO, and CAN, and collaborate with embedded engineers to define cross-boundary interfaces.
Integrate machine vision and camera systems, including image acquisition pipelines, sensor calibration, and routing vision outputs to downstream control and inference logic.
Collaborate with the ML Engineer to define inference APIs, data contracts, and performance budgets between model outputs and physical actuators.
Develop real-time and near-real-time control loops on Linux (PREEMPT_RT) and RTOS targets, with a clear understanding of scheduling, jitter, and determinism requirements.
Build hardware-in-the-loop (HIL) and integration test infrastructure that can verify system behavior with and without live hardware.
Document software architecture, interface contracts, timing assumptions, and integration procedures for both internal engineering and regulatory purposes.
Proactively identify integration risks, timing failures, and hardware/software boundary issues before they surface as field problems.