Physical AI Engineer

Civ Robotics • San Francisco, CA

2d • Onsite

About The Position

Civ Robotics is on a mission to automate repetitive tasks within the $3 trillion infrastructure construction industry. We’re dedicated to bridging the workforce gap and accelerating the development of essential infrastructure projects. At the core of our mission is our innovative technology in robotics and autonomous navigation. With over 100 robots deployed and operating across construction sites worldwide, we pride ourselves in ushering in the next generation of construction layout tools for solar, civil, land surveying, road striping, and general contracting. Our technology is engineered from the ground up to make our customers’ projects more precise, simple, and efficient than ever before. Join us and become an integral member of our dynamic team, leading the charge in pioneering construction robotics innovation! You’ll explore, prototype, and deploy new approaches to intelligence on real robots. That could mean: Building learning-based perception and vision systems for unstructured outdoor environments Exploring Vision-Language-Action models, foundation models, imitation learning, reinforcement learning, and learned world models Creating ways for robots to learn from demonstrations, operators, previous missions, and fleet data Combining classical robotics with modern machine learning rather than treating them as competing philosophies Developing models for scene understanding, terrain understanding, object detection, segmentation, localization, manipulation, and autonomous behavior Finding ways to turn large amounts of robot sensor data into useful training data Building data collection, evaluation, replay, and training pipelines for physical AI Taking recent research and figuring out what actually works on a robot Designing experiments, testing them in simulation, and then getting outside and putting them on a machine Working closely with autonomy, controls, embedded, mechanical, and product engineers Helping define the longer-term Physical AI architecture for our robots and fleet At Staff level, you'll also help identify what we should be working on next - not just execute an existing roadmap.

Requirements

• Deep experience in some combination of: Machine learning / deep learning, Computer vision, Robotics and autonomous systems.
• Proficiency in C++ and Python.
• Experience with PyTorch or similar ML frameworks.
• Experience with ROS / ROS2.
• Experience with real-world sensor data: cameras, LiDAR, GNSS, IMU, or similar.
• Experience training, evaluating, and deploying models on real systems.
• Ability to make things work outside the lab.
• Understanding of latency, compute, bad sensors, edge cases, changing environments, debugging, and the details between an idea and a working machine.

Nice To Haves

• Experience with Vision-Language Models / Vision-Language-Action models.
• Experience with Imitation learning or reinforcement learning.
• Experience with Robot foundation models.
• Experience with 3D perception.
• Experience with Neural rendering / NeRF / Gaussian Splatting.
• Experience with Self-supervised or unsupervised learning.
• Experience with Synthetic data and simulation.
• Experience with NVIDIA Jetson / TensorRT / edge inference.
• Experience with Isaac Sim or other robotics simulators.
• Experience with large-scale robotics datasets.
• Experience with Learning from teleoperation or human demonstrations.
• Experience in Heavy equipment, autonomous vehicles, agricultural robots, mining, construction, or other outdoor robotics.

Responsibilities

• Explore, prototype, and deploy new approaches to intelligence on real robots.
• Build learning-based perception and vision systems for unstructured outdoor environments.
• Explore Vision-Language-Action models, foundation models, imitation learning, reinforcement learning, and learned world models.
• Create ways for robots to learn from demonstrations, operators, previous missions, and fleet data.
• Combine classical robotics with modern machine learning.
• Develop models for scene understanding, terrain understanding, object detection, segmentation, localization, manipulation, and autonomous behavior.
• Find ways to turn large amounts of robot sensor data into useful training data.
• Build data collection, evaluation, replay, and training pipelines for physical AI.
• Take recent research and figure out what actually works on a robot.
• Design experiments, test them in simulation, and then get outside and put them on a machine.
• Work closely with autonomy, controls, embedded, mechanical, and product engineers.
• Help define the longer-term Physical AI architecture for our robots and fleet.
• At Staff level, help identify what we should be working on next.

Benefits

• Comprehensive healthcare coverage (medical, dental, vision) for you and your family.
• Competitive salary with growth potential.
• Equity options in a fast-growing robotics company.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: No Education Listed