Principal Robotics Engineer

Caterpillar Caterpillar · Industrial · Irving, TX

The Principal Robotics Engineer will design, build, and ship autonomy components for large off-road machines, covering perception, planning, control, and learning. This role involves taking work from prototype to production on real hardware, with a focus on the full autonomy stack and potentially deep expertise in AI, motion planning, perception, controls, or SLAM. Experience with C++, Python, ROS, and full-lifecycle product development in real-world robotics or autonomous vehicles is expected. The role also emphasizes MLOps and training infrastructure for ML-focused candidates.

What you'd actually do

  1. Participating in the processes of test review and analysis, test witnessing and certification of robotics software.
  2. Design, build, and ship autonomy components across perception, planning, control, and learning.
  3. Take work from prototype to production on real hardware.
  4. Bring depth in your specialty while contributing broadly across the stack.
  5. Work in small, outcome-owning pods staffed from our capability benches.

Skills

Required

  • Knowledge of software product design; ability to convert market requirements into the software product design.
  • Knowledge of the concepts, technologies and methodologies of artificial intelligence (AI); ability to develop, implement and/or apply artificial intelligence products and services in specific industry domain to achieve business goals.
  • Knowledge of relevant programming languages and tools; ability to test, write, design, debug, troubleshoot and maintain source codes and computer programs.
  • C++ and Python, production-quality.
  • ROS (or similar) in real robotics systems.
  • Full-lifecycle product experience — ideally has taken robotics products from idea to production, not just research or prototypes.
  • Pragmatic, first-principles engineer who owns components end-to-end and collaborates across a fast-moving delivery team.
  • Real-world / field robotics or autonomous vehicles (on- or off-road).
  • Learning-based control or perception, or foundation / VLA-style models.
  • Sim and on-vehicle deployment.
  • MLOps / training-infrastructure depth for the ML-leaning profile.

Nice to have

  • Deep expertise in at least one of: AI, motion planning, perception, controls, or SLAM.
  • Excited to get out on site and see their own code operate on iron — real machines in the real world, not just in the lab.

What the JD emphasized

  • production on hardware
  • full-lifecycle product experience
  • real-world / field robotics or autonomous vehicles
  • production-quality

Other signals

  • AI models
  • edge computing architectures
  • software systems that scale
  • new generation of machines that continuously learn
  • autonomy stack
  • perception through motion to control
  • modern learning-based methods
  • classical robotics foundations
  • full stack
  • prototype through to production on hardware
  • real-world / field robotics or autonomous vehicles
  • learning-based control or perception
  • foundation / VLA-style models
  • MLOps / training-infrastructure depth