Research Engineer - Generative Humanoid Motion Generation

Location: Zürich, Zurich, Switzerland

Department: AI Engineering

Workplace: on_site

Employment Type: full

Description

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 deployment of humanoids. We are founded by leading scientists in robot reinforcement learning (ex-Nvidia, ex-ETH Zürich) and backed by leading international VC firms. Within a few months, we’ve gone from our first line of code to deploying real humanoid capabilities with our partners.

The Role:

We are looking for an expert in whole-body motion generation and/or vision-conditioned generative models to strengthen our Zurich team.

Inspired by recent successes in neural avatars and computer graphics, we aim to apply similar principles to generate whole-body motions for our humanoid robots.

The goal of this position is to develop and deploy state-of-the-art multimodal generative models for robot trajectory generation.

Requirements

Hands-on experience in the development and training of generative models, including diffusion, flow matching, or masked-Transformer models. Ideally, this includes either training robotic imitation policies, training models to generate whole-body motions for human avatars, or training a vision-conditioned generative model to predict actions.

Particularly:

• PhD or master's degree in generative modeling, human motion generation, or learning-based robot trajectory generation with relevant project experience.
• Strong research profile with a high track record of publications at top computer vision, graphics, and/or robotic conferences, such as ICCV/ECCV, CVPR, SIGGRAPH, CORL, RSS, etc.
• Excellent knowledge of Python and PyTorch, with hands-on experience in training and fine-tuning generative models, including diffusion, flow matching, and masked-transformer models.
• Experience in deploying learning-based trajectory generation for robotic systems.
• Experience with modern GPU-based simulations such as Omniverse or Genesis.
• Good knowledge of state-of-the-art machine learning architectures.

Additionally, the following skills are a plus but not required:

• Experience with multi-modal generative models.
• Experience with image or video-based action prediction from ego-centric views.
• Experience with ego-centric world models for human avatars or humanoid robots.
• Experience with finetuning foundation models, e.g., Gr00t or SmolVLA, to produce whole-body actions or kinematics motions.
  

We are looking for a person who enjoys working in a team in a very dynamic and fast-moving environment, and who is able and willing to take ownership of projects and decisions.

Benefits

• Competitive compensation
• A front-row seat at one of Europe’s most ambitious robotics companies
• An energetic, collaborative team with a bias for action