Path Robotics
Machine Learning Engineer, Robot Learning, Loco-Manipulation
About this role
Path Robotics is seeking a Machine Learning Engineer to establish a new Robot Learning team focused on loco-manipulation in heavy manufacturing. This founding position involves designing the robot-learning stack and developing AI-driven systems that enhance robotic precision and adaptability in real-world applications.
What you'll do
- Develop the team's robot-learning stack from scratch.
- Establish ML infrastructure including training pipelines and data management.
- Train robotic action policies for manipulation and locomotion.
- Implement a phased rollout strategy for real-world deployment.
- Collaborate with engineers and domain experts across multiple disciplines.
What they're looking for
- Ph.D. or Master's in relevant field or equivalent experience
- 2+ years of robot learning experience
- Experience with sim-to-real transfer
- Familiarity with reinforcement learning for robotics
- Strong Python programming skills
- Hands-on experience with simulation tools like NVIDIA Isaac Sim
- Ability to work with physical robots
- Effective communication skills
Benefits
- Daily free lunch
- Flexible PTO
- Comprehensive health coverage
- Paid parental leave (6 weeks fully paid for all, up to 14 weeks for birthing parents)
- 401(k) retirement plan
- Employee referral bonuses
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
Path Robotics
Path Robotics develops autonomous robotic welding systems with adaptive motion planning and AI-driven capabilities for manufacturing. The company is hiring welding engineers, mechanical engineers, machine learning engineers, and technical marketing engineers to advance its mobile robotic welding solutions.
- Website
- pathrobotics.com
Likely interview questions
- Walk us through a robot learning project where you trained a policy in simulation and deployed it on real hardware. What broke during sim-to-real transfer, and how did you debug it?
- Describe your hands-on experience with diffusion-based or flow-matching action policies for robots. How did you handle action chunking, and what challenges did you encounter?