Machine Learning Research Engineer/Scientist

What to Expect

In building robots for the home, we are tackling a true grand challenge in robotics: dexterous and safe mobile manipulation in unstructured environments. To make this possible, we are building across the entire robotics stack. We’re training state-of-the-art AI models that leverage our large-scale, high-quality, real-world data collection system.

At the same time, we’re building a new kind of consumer hardware product, which will be deployed into homes and delivering real value for real customers. As an early member of a small, cross-functional team, you’ll play a key role in pushing both our technology and product forward: advancing the frontier of embodied AI, and soon giving countless hours back to our customers so they can spend more time on the things they value most.

As a Machine Learning Research Engineer, you will work on the software and algorithms that enable our robots to complete dexterous manipulation tasks in home environments. You will leverage our unique large-scale in-the-wild data collection operation and our growing robot fleet to continuously add new behaviors and improve robustness across tasks and environments.

What You'll Do

• Design and develop state-of-the-art robot learning algorithms for manipulation and controls
• Leverage large-scale in-the-wild data collection to develop generalizable robot behaviors
• Own the end-to-end loop from task definition and data curation to model training, evaluation, and on-robot deployment
• Collaborate closely with a full-stack robotics team to develop a consumer personal robot
• Write maintainable, production quality code for research and deployment

What You'll Bring

• 3+ years (or equivalent) working on machine learning for robotics, controls, or perception
• Proficiency with Python and any deep learning framework (PyTorch preferred)
• Hands-on experience with robot learning for real-world tasks, including offline/online training and evaluation
• Implement tooling for visualization, debugging, and evaluation of data and learning pipelines
• Communicates clearly and collaborates effectively, thriving in fast-moving, cross-functional teams

Nice to Have

• Familiarity with classical robotics fundamentals (e.g. controls, planning, state estimation)
• Comfortable profiling and optimizing for latency, memory, and compute on edge devices
• Experience with large-scale model training pipelines (LLM/VLM/VLA)