AI Researcher - Perception

We're building humanoid robots that work in home - doing the chores, handling the tasks, and giving people their time back. Simple, but it's not.

To do this right, we have to solve robotics, AI, manufacturing - at the same time, at scale, in a form factor that has to be safe enough to live with your family. If you're inspired by this, you'll thrive here. We've been at this since 2014 and we're at the point where the hard problems are behind us and the hard work is in front of us.

NEO is our flagship - a home robot designed to move, learn, and operate in the real world alongside real people. We're not demoing it - we're shipping it. We're excited to meet you, if this excites you.

If you've spent your career working on problems that matter and want to see them actually reach the world - this is that moment. We're scaling, we're hiring with intention, and we need people who want to build something that will genuinely change how humans spend their time - safely creating abundance for all.

The Motion team enables NEO to move through and interact with the world. We build the perception NEO needs to understand its surroundings and locomote through any environment, and the control that lets it use its whole body to accomplish real tasks - crawling, bracing, climbing, lifting with more than just its arms. Because NEO operates around people, safe and compliant motion is a design constraint on everything we ship, not a feature layered on top.

Own perception for NEO from first prototype to fleet-wide deployment. You will build the vision systems that measure the robot's own geometry, recover human and object pose, and reconstruct the terrain and scenes NEO has to walk through and work in. How precisely NEO places a hand or plans a step is bounded by the perception you deliver, and you will measure your impact on deployed robots rather than on benchmarks.

• Kinematic Calibration: Recover each robot's true geometry from camera data so controls and learned policies act on an accurate model of the body they drive.

• 3D Scene Understanding: Deliver depth and scene reconstruction accurate enough to drive terrain-aware locomotion and contact-rich manipulation.

• Pose Estimation: Ship human & object pose estimation models that can be used for large scale data labelling efforts and human/robot motion generation models.

• Sensor Fusion: Drive time synchronization and fusion across cameras and inertial sensors so every downstream consumer gets spatially and temporally consistent data.

• Calibration at Scale: Own intrinsic and extrinsic camera calibration - automated on the production line, monitored for drift across the fleet, and recoverable through field tools.

• Evaluation: Establish the datasets and metrics that show whether a perception change improved robot behavior, not just an offline score.

• Geometric Vision: Fluent in camera models, multi-view geometry, and non-linear optimization; reaches for classical or learned methods based on the problem.

• Production ML: Trains and ships modern vision models, including transformer backbones, generative approaches, and learned depth, and makes them fast and predictable on embedded compute.

• Hardware Grounding: Treats sensor noise, timing, and calibration drift as first-class problems, and traces how each propagates into a control loop or policy rollout.

• Prototype to Production: Stands up a new capability from a paper and a test rig, then makes it repeatable across a fleet and monitored in the field.

• Technical Direction: Sets the approach for a problem area, defends tradeoffs across sensing, models, and compute budget, and stays hands-on.

• Proven track record of shipping perception systems that run on physical hardware in the real world - robotics experience welcome but not required

• Demonstrated ability to train, evaluate, and deploy deep learning models for detection, depth, or pose estimation in PyTorch

• Deep experience with geometric computer vision, camera-based kinematic calibration (hand-eye / foot-eye calibration), and camera calibration including intrinsics & extrinsics

• Experience owning a capability end to end, from problem definition through deployment and iteration on real failures

• Strong Python and modern C++ in large shared codebases, with solid testing and code review practices

• 5+ years of relevant industry experience, or an advanced degree plus equivalent hands-on experience

• Experience with visual-inertial odometry or SLAM, including multi-sensor time synchronization and fusion

• Familiarity with robot kinematics and frame transforms, and how calibration error shows up in manipulation and locomotion

• Experience building and adapting segmentation and classification models, including fine-tuning open-vocabulary or foundation backbones for a specific domain

• Experience reconstructing scenes and terrain from video using differentiable rendering, neural rendering, or Gaussian splatting

• Exposure to automating calibration and validation in manufacturing, or to building field diagnostic tools for deployed hardware

• Familiarity with sensing beyond RGB such as radar, thermal, or near-infrared

• Salary Range: $200,000 - $300,000 + Equity

• Comprehensive medical, dental, and vision coverage

• Generous paid time off, company holidays, and parental leave

• 401(k) plan with company match (100% on the first 3% of contributions, 50% on the next 2%)

• Flexible Spending Accounts (FSA) and Health Savings Accounts (HSA) options

• Commuter benefits (transit and parking)

• Short-term and long-term disability, and life insurance

• Employee Assistance Program (EAP) for mental health, financial, and personal support

• Onsite snacks and catered lunches