Software Engineer, Robot Autonomy (Actuator Control & Locomotion), Intern

About the role

As a Reinforcement Learning Intern, you will train locomotion and low-level control policies for legged security robots, transitioning them from simulation to real-world hardware. You will work closely with actuators to ensure robots can navigate varied terrain and challenging weather conditions. This role involves designing training setups, running experiments, and measuring policy performance in both simulated and physical environments.

Responsibilities

• Train and evaluate reinforcement learning policies for locomotion and low-level control in simulation.
• Apply sim-to-real techniques such as domain randomization, reward design, and policy robustness methods.
• Test and validate policy performance on physical robots across varied terrain and weather conditions.
• Explore control approaches that transfer across different robot embodiments and dynamics.
• Build evaluation workflows with clear metrics and repeatable experiments.
• Apply solid engineering practices including experiment tracking, version control, and reproducible training runs.

Required skills

• Reinforcement learning
• Robot dynamics
• Control theory
• Python
• PyTorch
• JAX
• Docker
• Git
• Isaac Sim
• MuJoCo
• Gazebo

Nice to have

• Physical robot deployment
• Legged robotics
• Actuator control
• Motor control
• Joint control
• C++
• ROS 2

Qualifications

• Currently pursuing a PhD in Robotics, Machine Learning, Computer Science, or a related field
• Recently completed a Master's degree in Robotics, Machine Learning, Computer Science, or a related field

Benefits

• Equity package
• Career growth opportunities

About the Company

Laelaps AI is a startup based in Zurich building intelligent software for physical security robots. The company is backed by visionary investors and led by a team of PhD-level co-founders in AI, Robotics, and Physics.