Research Engineer, Robot Learning

present of AI; San Carlos (CA); On-site; USD 120,000 - 160,000; Full time

Job summary

1X designs and manufactures bipedal humanoid robots engineered to execute household chores and personal assistance tasks safely, according to company statements. This role focuses on building and improving training systems for reinforcement learning that teach robots to move and handle tasks, transferring policies from simulation to real hardware.

A master’s degree background, about a year of hands-on robotics and ML work, solid physics modeling, and proficiency in Python for training and C++

Qualifications

• Master's-level background required.
• About 1 year of hands-on robotics and ML work.
• Solid physics modeling and simulation experience.

Responsibilities

• Will build and improve training systems for reinforcement learning of humanoid robots.
• Transfer learned policies from simulation to real hardware with a focus on safety and reliability.
• Diagnose and fix sim-to-real gaps to improve real robot behavior.

Skills

Reinforcement learning

Robotics

Python

C++

Sim-to-real

Physics modeling

Education

Master's degree in Robotics/ML/CS

Tools

Python

C++

Robotics simulation

Open-source ML frameworks

Job description

This role is about building and improving the training systems that teach humanoid robots to move and handle tasks using reinforcement learning. It asks for a master's level background, about a year of hands-on robotics and machine learning work, solid physics modeling, and comfort writing both Python for training and C++ for real-time robot software. A strong candidate can find why a simulated robot behaves differently from the real one and fix it.

How it fits the AI era

Learned policies are replacing hand-coded control for whole-body robot skills, so much of the work is shaping data, simulations and training runs rather than writing rules for each movement. AI speeds up how fast new behaviors can be trained, but people still decide what the simulation must capture, judge whether a result is safe, and trace faults when a robot fails on real hardware.

How it builds AI experience

This role gives direct experience training reinforcement learning policies and moving them from simulation onto physical humanoids meant for homes, which is a rare and valuable mix. Someone building toward it can train a legged or humanoid policy in an open simulator, use human motion capture data to guide it, and document how they narrowed the gap to real behavior.

Worth showing

• Sim2real reinforcement learning projects
• Rigid body dynamics and contact modeling
• Production Python for ML pipelines
• Real-time C++ on robots
• Human motion retargeting and kinematics

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About 1X

1X designs and manufactures bipedal humanoid robots engineered to execute household chores and personal assistance tasks safely, according to company statements.