Machine Learning Engineer

est. by Gemini 2.5 FlashCHF 120'000 – 140'000/ year

Your Mission

Develop, train, and iterate on reinforcement learning policies for humanoid locomotion and manipulation
Analyze training metrics and experiment results from simulation and real robots, identify issues, and propose follow-up actions
Investigate sim-to-real gaps and develop approaches to improve the transfer of learned policies to real robots
Develop machine learning solutions in Python using frameworks such as PyTorch, TensorFlow, or JAX
Build and maintain RL/IL training pipelines, including environments, curricula, normalization, and evaluation
Evaluate learning algorithms and adapt them to the specific problem at hand, with support from experienced team members
Work collaboratively in a multicultural and interdisciplinary environment on a novel humanoid robot
Proactively identify opportunities for improvement and contribute ideas to the team

PhD/MSc in Robotics, Computer Science, or a related technical field
Python programming skills and experience with machine learning frameworks such as PyTorch, TensorFlow, or JAX
Ability to independently write and debug basic C++ code
Hands-on experience developing and training reinforcement learning policies
Experience with imitation learning or related learning-based methods is a plus
Ability to interpret training metrics and experimental results and use them to guide the next iteration
Understanding of sim-to-real challenges in robotics and approaches to address them
Autonomous and proactive working style, with a strong drive to experiment, learn, and solve problem

Experience working with real robots
Experience with modern RL/IL/ML approaches, such as transformers or cross-embodiment transfer learning
Experience with Isaac Sim or other standard RL simulators
Experience working independently with ROS2, including writing nodes and debugging interactions between them
Decent C++ programming skills
Experience defining objectives for your work and integrating them into a broader team plan or owning an epic