Deep Learning Robot Manipulation Engineer

This company is developing and commercializing rugged, multi-purpose humanoid robots designed to perform real, economically valuable work in real-world environments. The founding team brings decades of experience across humanoid robotics, bionics, and large-scale hardware product development, with a track record of delivering systems that have operated in extreme environments, including space, deep-sea, and high-traffic public settings.

The mission is centered on shipping beautiful, reliable robotic products at scale, with a strong emphasis on hardware robustness, real-world deployment, and customer-driven engineering. The team is building foundational robotics technology with the intent to move beyond demos and into production-grade systems.

Anonymised Role Summary: Deep Learning Manipulation Engineer

This is an early, high-impact role focused on enabling dexterous manipulation for humanoid robots operating in unstructured environments. The engineer will help define and build the company’s deep learning manipulation strategy from the ground up, working on high-DOF robotic hands and full-body manipulation systems.

The role blends cutting-edge ML research with real-world deployment, emphasizing models that actually run on hardware, integrate with sensing and control systems, and perform reliably outside the lab. Candidates with experience shipping products or production ML systems are especially valued, though strong research backgrounds with demonstrated execution are also considered.

Core Responsibilities

• Design and train deep learning models for dexterous robotic manipulation, including grasping, tool use, long-horizon tasks, and in-hand object manipulation.

• Apply curriculum learning approaches to scale from simple interactions to complex, multi-step behaviors.

• Integrate tactile sensing and proprioception into closed-loop, end-to-end learning pipelines.

• Collaborate with teleoperation and data teams on data collection, labeling, versioning, and scaling strategies.

• Adapt and extend state-of-the-art manipulation approaches such as behavior cloning, diffusion policies, and foundation models.

• Deploy trained models onto real humanoid hardware with attention to latency, safety, and system integration.

• Build and maintain evaluation pipelines across simulation and real-world testing to measure robustness and generalization.

• Contribute to the growth and technical direction of the ML and autonomy teams.

Ideal Background

• Advanced degree in Robotics, Computer Science, or a related field.

• Several years of hands-on experience applying deep learning to robotic manipulation.

• Strong grounding in modern manipulation methods, including vision-language-action models and large-scale learning approaches.

• Experience working with large datasets, cloud infrastructure, and production ML systems.

• Proven ability to write high-quality, maintainable software.

• Comfort operating in fast-moving, ambiguous startup environments with high ownership.

Bonus Signals

• Experience across broader robotics perception stacks such as vision, point clouds, and object detection.

• Publications at top-tier ML or robotics conferences.

• Direct experience deploying robots and training models that operate reliably in production settings.