State Estimation Engineer - Data Collection Systems

Figure is an AI Robotics company autonomous general-purpose humanoid robots. The goal of the company is to ship humanoid robots with human level intelligence. Its robots are engineered to perform a variety of tasks in the home and commercial markets. We are based in North San Jose, CA and require 5 days/week in-office collaboration. It’s time to build.

We are looking for a State Estimation Engineer to own the architecture, algorithm development, and calibration workflows for a next-generation data collection system. This system powers two core capabilities: low-latency real-time teleoperation of our humanoid robots and ultra-high-precision offline trajectory reconstruction for data collection and policy training. You will build and deploy dual-tier estimation pipelines and user-onboarding calibration routines that fuse heterogeneous sensor modalities to track full-body human kinematics and floating-base motion across dynamic tasks.

Key Responsibilities:

• Design and implement dual-tier state estimation algorithms in modern C++: low-latency, real-time filters for streaming teleoperation and batch optimization/smoothing routines for high-accuracy offline dataset generation.
• Own and develop subject-calibration procedures, designing rapid, intuitive routines to estimate individual body segment dimensions, joint offsets, and sensor-to-body extrinsics whenever a user equips the system.
• Develop robust sensor fusion architectures combining spatial transforms, visual-inertial data, and inertial signals into full-body kinematic pose estimates.
• Address spatiotemporal sensor calibration, dynamic environmental interference, and kinematic constraint enforcement on human skeletal models.
• Develop techniques to extract useful information from compliant tactile sensing in the presence of large sensor deformation, stretching or folding.
• Diagnose and understand limitations of existing hardware or designs and inform future design requirements.
• Evaluate novel sensing modalities to inform future hardware designs.
• Build diagnostic tooling, validation pipelines, and error analysis workflows to evaluate accuracy for both online and offline algorithms.

Requirements:

• 4+ years of experience building multi-sensor fusion and state estimation solutions for dynamic hardware systems.
• Hands-on expertise with both real-time filtering techniques ((E)KFs, sliding-window estimators) and offline batch optimization tools (Factor Graphs, GTSAM, Ceres, Non-Linear Least Squares).
• Proven capability to design fast, reliable calibration, zeroing, and alignment workflows for multi-sensor suites and kinematic models.
• Deep mathematical foundation in 3D spatial kinematics, Lie groups (SE(3), SO(3)), forward/inverse kinematics, and constrained optimization.
• Proven ability to write high-performance, modular C++ for embedded or edge computing platforms alongside Python for data analysis and visualization.

Bonus Qualifications:

• Experience with low-latency streaming pipelines for teleoperation, haptics, or human-in-the-loop control systems.
• Background in human biomechanics, skeletal tracking, or body-mounted telemetry systems.
• Prior experience applying Machine Learning (ML) techniques to motion priors, trajectory smoothing, or learned state estimation/calibration.

The US base salary range for this full-time position is between $150,000 and $300,000 annually.

The pay offered for this position may vary based on several individual factors, including job-related knowledge, skills, and experience. The total compensation package may also include additional components/benefits depending on the specific role. This information will be shared if an employment offer is extended.