Dexterous Manipulation Research Engineer (VLA)

Flexion Robotics; San Francisco (CA); Vor Ort; USD 180.000 - 260.000; Vollzeit

Benefits dieser Stelle

Competitive compensation

Enhanced pension plan

Enhanced holiday & paid leave perks

Relocation sponsorship

Central Zürich office facilities

Zusammenfassung

Flexion is seeking an expert in dexterous manipulation to lead the development of physical foundation models using egocentric video to enable humanoid robots to interact with the world with human-like fluidity. You will own model architecture, integrate egocentric priors with real-time robot policies, and ensure reliable manipulation in unstructured environments.

The role requires PhD or Master’s in Robotics/ML, hands-on experience with Humanoid or Dexterous Manipulation, and deep knowledge of

Qualifikationen

• PhD or Master’s in Robotics or ML with focus on data-driven manipulation, egocentric vision, or foundation models.
• Experience with Humanoid or Dexterous Manipulation and contact-rich physics.
• Strong Python, PyTorch, and large-scale training knowledge (FSDP, NCCL).
• Expertise in Diffusion Models, Flow Matching, and Transformers.

Aufgaben

• Architect and implement scalable egocentric pre-training objectives for hand-object interactions.
• Develop and scale Vision-Language-Action foundation models for robotic trajectories.
• Design and optimize generative action heads using diffusion and flow-matching techniques.
• Align human-centric video representations with 20+ DoF humanoid constraints for stable control.
• Utilize offline RL and high-fidelity simulation to refine model performance.
• Translate cutting-edge research into production-ready architectures for reliability.

Kenntnisse

Robotics research

Hands-on experimentation

Problem solving

Team collaboration

Ausbildung

PhD or Master's in Robotics or ML

Tools

Python

PyTorch

Diffusion Models

Flow Matching

Transformers

MuJoCo

IsaacGym

Distributed training (FSDP, NCCL)

Jobbeschreibung

Flexion is seeking an expert in dexterous manipulation to lead the development of physical foundation models using egocentric video to enable humanoid robots to interact with the world with human-like fluidity. You will own model architecture, integrate egocentric priors with real-time robot policies, and ensure reliable manipulation in unstructured environments.

The role requires PhD or Master’s in Robotics/ML, hands-on experience with Humanoid or Dexterous Manipulation, and deep knowledge of