Responsibilities

Design, implement, and deploy state estimation and sensor fusion algorithms for real-time general-purpose robot control — EKFs, UKFs, particle filters, factor graphs — fusing IMUs, encoders, force/torque sensors, and proprioceptive signals
Develop and tune advanced control algorithms for dynamic robot motion: nonlinear control, model predictive control (MPC), optimal control, and whole-body control for legged and manipulating systems
Architect and ship production-grade C++ control code running in real-time embedded environments; hold your implementations to the same quality bar as deployed software
Iterate rapidly between simulation and hardware — design experiments, collect data, debug failure modes, and drive measurable performance improvements on physical robots
Develop trajectory optimization and motion planning algorithms that respect actuator limits, contact constraints, and stability margins
Define and maintain performance metrics and evaluation frameworks for control and estimation subsystems; own the failure analysis loop
Work directly with embedded, mechanical, and AI teams to integrate control algorithms across the full robot stack

Minimum Qualifications

5+ years of professional experience developing control systems for dynamic robots, deployed on real hardware
Master's or PhD in Robotics, Controls, Mechanical Engineering, or related field
Deep expertise in control theory: nonlinear control, MPC, LQR, optimal control, and whole-body control
Strong state estimation background: Kalman filters (EKF/UKF), particle filters, factor graphs, and Bayesian estimation
Production-quality C++ for real-time control; Python for analysis, simulation, and tooling
Solid command of robot kinematics, rigid-body dynamics, and spatial mathematics
Hands-on experience with sensor integration and characterization: IMUs, encoders, force/torque sensors
Proven track record implementing and validating control algorithms on physical robotic systems — not just simulation

Experience with bipedal, quadruped, or humanoid robots — highly dynamic, underactuated, contact-rich systems
Background in reinforcement learning or learning-augmented control for legged locomotion or manipulation
Experience with whole-body control and contact dynamics: contact estimation, impact modeling, friction-cone constraints
Familiarity with trajectory optimization frameworks and solvers: OSQP, IPOPT, Crocoddyl, or custom implementations
Proficiency with simulation environments: MuJoCo, Drake, Isaac Sim, or equivalent
Experience with real-time computing constraints: deterministic execution, latency budgets, and embedded deployment
Track record of publications at top-tier venues (ICRA, IROS, CoRL, RSS, IJRR) is a strong plus