Head of Robot Capabilities & Learning

Praxy; San Diego, Northern (CA, KY); Hybrid; USD 180,000 - 280,000; Full time

Benefits offered by this job

Health coverage

23 days PTO and sick leave

401(k) with employer match

Equity stock options

Free daily catered lunch

Job summary

Humanoid is seeking a Head of Capability Development to lead a team of robot learning engineers, delivering reliable locomanipulation capabilities on real hardware. The role blends applied deep learning with hands-on debugging and data curation, reporting to leadership on progress and failures.

You will own the roadmap, hire and coach engineers, and collaborate across Control, Core Technology, and Simulation teams to push a capable humanoid platform toward real-world deployment.

Qualifications

• 5+ years working on robots with shipped artifacts.
• Experience leading a team of 5+ engineers doing applied robot learning.
• Hands-on today: trained and debugged policies within the last year and can curate data.
• Experience post-training VLA-class policies via behavior cloning and/or RL.
• Solid understanding of whole-body control, teleoperation, and RL locomotion controllers.
• Familiarity with deep learning infrastructure: streaming datasets, checkpointing, distributed training (PyTorch or JAX).
• Strong software engineering practices and clear documentation of experiments.

Responsibilities

• Own The Capability Roadmap: define use-case priority and evaluation plan.
• Lead The Team: hire, develop, and manage robot learning engineers; set standards.
• Co-Own The System 0/1 Interface: align with Controls on policy commands and debug boundary issues.
• Partner On Data Strategy: specify data needs and oversee collection for bipedal locomanipulation.
• Drive Sim-To-Real Transfer: coordinate RL training on the digital twin and improve transfer to real robots.
• Shape Base Model Development: integrate embodiment coverage into pretraining and adopt new models.
• Inform Hardware Design: feed findings to future hardware iterations.
• Report On Progress: communicate capabilities, eval results, and failures to leadership.

Skills

Robot learning leadership

People management

Hands-on policy debugging

Deep learning (PyTorch/JAX)

RL & locomotion controllers

System integration

Tools

PyTorch

JAX

RL frameworks

Simulation tools

Job description

Humanoid is seeking a Head of Capability Development to lead a team of robot learning engineers, delivering reliable locomanipulation capabilities on real hardware. The role blends applied deep learning with hands-on debugging and data curation, reporting to leadership on progress and failures.

You will own the roadmap, hire and coach engineers, and collaborate across Control, Core Technology, and Simulation teams to push a capable humanoid platform toward real-world deployment.