Robotics Software Engineer, Autonomy & Behavior

Anyware Robotics•Fremont, CA

About The Position

Anyware Robotics builds general-purpose mobile manipulator robots for industrial applications. Our robots are trusted by customers across logistics, retail, and manufacturing, supporting applications such as truck unloading, mobile palletizing, and machine tending. The robots run on AnywareOS, our industrial physical intelligence system. Its architecture is modular and designed to generalize across applications without reprogramming. We are looking for a Robotics Software Engineer to own the behavior layer of AnywareOS — the task sequencing, fault recovery, and cross-component interfaces that turn planning, perception, and controls into a working production cycle. It is where learned policies meet deterministic guarantees — and what a new capability has to clear before it runs on a customer floor.

Requirements

2+ years of professional experience building robotics software that has run on physical hardware in production.
Strong C++ and Python, including in codebases you did not write.
Ownership of a feature end to end, through design, implementation, testing, and deployment.
Demonstrated ability to debug across component boundaries from logs and telemetry.
Strong fundamentals in spatial transforms, robot kinematics, and 3D geometry.
Depth in robot task sequencing and failure recovery — behavior trees, task planners, or a custom sequencer — plus at least one of: motion planning, perception, or controls.

Nice To Haves

ROS 2 and production robot deployment pipelines.
Warehouse, logistics, or industrial automation experience.
Replay, telemetry, or regression infrastructure for physical systems.
Learned components shipped into production behind deterministic safeguards, and judgment about where their output can and cannot be trusted.
Broad ownership across the stack at an early-stage company.

Responsibilities

Design the robot's behavior and recovery logic — how Pixmo sequences a task, detects that a cycle has failed, and restores the system to a defined state from which the cycle can resume without operator intervention.
Improve cycle time and task success rate on production applications including trailer unloading, palletizing, and depalletizing.
Build the instrumentation the system is diagnosed through — logging, replay, and simulation tooling capable of reproducing intermittent field failures.
Take field failures to root cause — from a telemetry trace of something that happened once, on one robot, at a customer site, to the specific defect, the fix, and the regression test that keeps it from coming back.
Turn new capabilities into production behavior — when planning, perception, or controls lands something new, you are the engineer who decides what it takes to trust it on a customer floor.