Lead - Dexterous Manipulation

About Flexion:

At Flexion, we're building the intelligence layer powering the next generation of humanoid robots. Our mission is to accelerate the transition from fragile prototypes to real-world humanoid deployment. We are founded by leading scientists in robot reinforcement learning (ex-Nvidia, ex-ETH Zürich), and backed by leading international VC firms. In just months, we’ve gone from our first line of code to deploying real humanoid capabilities.

The Role:

We are seeking a Manipulation Lead to define and drive Flexion's manipulation stack, with a strong focus on learning-based dexterous control for humanoid robots.

In this role, you will own the technical direction, architecture, and delivery of manipulation capabilities, from research ideas to real-time execution on physical robots.

You will work closely with the perception, controls, and infrastructure teams, and lead the engineers working on dexterous manipulation. This is a hands-on technical leadership role, not a purely managerial position.

Responsibilities

• Own the full manipulation stack, from problem formulation and data collection to training, deployment, and evaluation on real robots.
• Define the technical roadmap for dexterous manipulation.
• Lead the transition from research prototypes to robust, real-time controllers running on humanoid hardware.
• Design, implement, and deploy learning-based manipulation controllers (e.g. diffusion-based policies, flow matching, RL, or hybrids).
• Establish best practices for training, sim-to-real transfer, evaluation metrics, and debugging on hardware.
• Mentor and guide the engineers working on manipulation.
• Collaborate closely with perception, whole-body control, and infrastructure teams to enable closed-loop manipulation.

Requirements

• PhD degree in Robotics, Machine Learning, or a related field, with significant hands-on experience in learning-based manipulation.

• Deep expertise in dexterous manipulation models, such as:
• Diffusion models
• Flow matching
• Reinforcement learning
• Strong understanding of robot control and real-time constraints.
• Proven experience deploying learning-based controllers on real robotic hardware.
• Excellent proficiency in Python and PyTorch, including training large neural networks.
• Solid knowledge of transformers and modern generative models.
• Ability to take technical ownership and make high-impact decisions in a fast-moving environment.

Benefits

• Key technical leadership role in a young, fast-growing robotics company in Zurich.
• Direct influence on the core manipulation capabilities of next-generation humanoid robots.
• High ownership and autonomy in shaping systems and research direction.
• Competitive compensation, including base salary and additional benefits.