Our client is building rugged,
multi-purpose humanoid robots
designed to perform real work in real-world environments.
The founding team brings decades of experience across humanoid robotics, bionics, and large-scale hardware product development, with systems that have operated in extreme environments across land, sea, space, and complex public settings. Their mission is to ship beautiful, reliable robotic products at scale while building a highly customer-focused engineering team.
We are looking for a
Deep Learning Manipulation Engineer
to help train humanoid robots to perform dexterous manipulation tasks in the real world. This is an opportunity for someone who thrives on hard problems, enjoys ambiguity, and wants to see their work directly translated into deployed robotic systems. Candidates with experience shipping real products are especially compelling, though strong aptitude and drive are equally valued.
As one of the early hires in this team, you will help shape the overall deep learning and manipulation strategy, infrastructure, and technical direction. You’ll work on cutting-edge problems at the intersection of learning, control, sensing, and hardware.
What You’ll Do
- Design and implement advanced deep learning models and training pipelines for dexterous manipulation on humanoid robots with high degrees of freedom and multi-fingered hands
- Train models using curriculum learning approaches, progressing from simple interactions to precise grasping, long-horizon tasks, tool use, and in-hand manipulation
- Integrate tactile sensing and proprioception into end-to-end, closed-loop learning systems
- Partner closely with teleoperation and data teams to design scalable data collection, labeling, and versioning strategies
- Leverage state-of-the-art manipulation models while contributing new architectures for increasingly complex tasks
- Deploy learned policies onto robotic hardware with a focus on real-time performance, safety, and robust integration with control systems and sensors
- Collaborate on the design and optimization of the end-to-end manipulation ML pipeline
- Stay current with the latest research and industry developments in robotic manipulation and learning
- Build evaluation frameworks to rigorously test learned policies in simulation and real-world trials, measuring robustness and generalization
- Support the growth and development of the machine learning and autonomy teams
Requirements
- Grit and curiosity to tackle some of the hardest problems in robot manipulation
- Comfort working in fast-paced, ambiguous startup environments
- Master’s or PhD in Robotics, Computer Science, or a related field
- 3+ years of experience applying deep learning to robotic manipulation
- Strong grounding in modern approaches such as behavior cloning, vision-language-action models, diffusion policies, and foundation models
- Experience working with large-scale datasets and cloud-based training infrastructure
- Clear understanding of the challenges involved in deploying neural networks on real robotic systems
- Strong software engineering skills and a first-principles mindset
Nice to Have
- Experience with adjacent ML areas in robotics such as perception, point clouds, segmentation, and object detection
- Publications at top machine learning or robotics conferences
- Hands-on experience deploying robots, collecting large datasets, and training models that run reliably in production environments
Why Join
- Opportunity to shape the future of humanoid robotics and real-world autonomy
- Work alongside deeply experienced, mission-driven engineers and researchers
- Access to advanced labs, prototyping tools, and the freedom to experiment
- Competitive compensation, strong benefits, flexible working arrangements, and meaningful equity