Senior AI Software Engineer, Reinforcement Learning

What the role actually is

This is a robot-learning role on Agility's AI Controls team. The person would train, evaluate, and ship reinforcement-learning policies that let Digit move safely and reliably in production environments.

What you would work on

• Train RL policies for locomotion, manipulation, whole-body control, and dynamic interaction.
• Integrate perception into policies for obstacle-aware motion and perceptive manipulation.
• Build scalable training pipelines, simulation tasks, and evaluation infrastructure.
• Work with on-robot software and deployment teams to get learned policies onto deployed humanoids.

What they are asking for

• Experience developing and deploying RL policies for robotics applications.
• Strong Python and deep-learning framework skills.
• Familiarity with reward design, hyperparameter tuning, and exploration strategies for control tasks.
• Comfort turning simulation and evaluation work into production robot behavior.

Why this one is worth a look

Agility is one of the few humanoid companies with robots already running in customer facilities. The role is engineering-heavy, but it is directly about learned behavior for robots operating around people, which makes it relevant for anyone tracking the practical edge of robot learning.