Reinforcement Learning Engineer (Full-Time) - Humanoid Robot

Job Description

<h3>Job Description</h3><p>Job Description<p><strong>About AXIBO</strong><p>AXIBO is a robotics company pioneering the <strong>design, prototyping, and manufacturing</strong> of advanced robotic systemsall under one roof. We build everything in-house and take pride in delivering <strong>robust, reliable products</strong> that power automation across industries. Our fast-paced environment demands high levels of <strong>precision, organization, and execution</strong>not just in engineering, but across all functions.</p><strong>Position Overview</strong><p>As a <strong>Reinforcement Learning Engineer</strong>, you will develop and deploy machine learning systems that enable intelligent behaviors in our humanoid and legged robots. You&#39;ll work at the intersection of control theory, deep learning, and roboticshelping close the loop between simulation and reality to bring adaptive behaviors into real-world machines.</p><strong>Key Responsibilities</strong><ul><li><p><strong>Develop reinforcement learning agents</strong> for robotic control tasks such as locomotion, manipulation, and dynamic balance</p></li><li><p><strong>Implement learning architectures</strong> using policy gradient methods, actor-critic frameworks, and off-policy algorithms (e.g., PPO, SAC, TD3)</p></li><li><p><strong>Build reward functions</strong>, curriculum learning strategies, and simulation environments tailored for real-world transfer</p></li><li><p>Design <strong>multi-agent training pipelines</strong>, including distributed rollouts, experience replay, and adaptive difficulty scaling</p></li><li><p>Interface with <strong>Isaac Gym, Mujoco, Brax, and custom physics simulators</strong> to run large-scale experiments</p></li><li><p>Work with hardware and firmware teams to deploy trained policies to <strong>embedded or real-time environments</strong></p></li><li><p>Design diagnostic tools and visualization dashboards to monitor training progress and system behavior</p></li><li><p>Apply <strong>domain randomization, sim2real techniques</strong>, and sensor noise modeling to enhance policy robustness</p></li><li><p>Maintain code quality through <strong>version control, testing, and modular design</strong></p></li><li><p>Stay current with academic literature and integrate novel RL methods as appropriate</p></li></ul><strong>Required Skills and Qualifications</strong><ul><li><p>Bachelor&#39;s or Master&#39;s degree in Computer Science, Engineering, Robotics, or a related field</p></li><li><p><strong>2+ years of hands-on experience</strong> applying deep reinforcement learning to simulation or robotic control tasks</p></li><li><p>Strong grasp of <strong>machine learning fundamentals</strong> and <strong>control theory</strong></p></li><li><p>Proficiency with <strong>PyTorch</strong>, <strong>JAX</strong>, or <strong>TensorFlow</strong></p></li><li><p>Programming experience in <strong>Python</strong> and <strong>C++</strong></p></li><li><p>Deep understanding of <strong>policy optimization</strong>, generalization, and environment design</p></li><li><p>Experience working in <strong>Linux</strong> development environments and with <strong>GPU-based training</strong> pipelines</p></li><li><p>Excellent debugging skills across ML, software, and hardware stacks</p></li><li><p>Ability to independently manage experiments and rapidly iterate on model architectures</p></li></ul><strong>Preferred Experience (Bonus)</strong><ul><li><p>Deployment of RL systems to <strong>real-world robots</strong>, especially legged or humanoid platforms</p></li><li><p>Contributions to open-source RL frameworks or robotics middleware (e.g., ROS, Isaac ROS)</p></li><li><p>Experience with <strong>imitation learning</strong>, <strong>behavior cloning</strong>, or <strong>inverse reinforcement learning</strong></p></li><li><p>Prior <span >research/publications</span> in reinforcement learning, multi-agent systems, or robotic control</p></li><li><p>Familiarity with <strong>low-level robot interfaces</strong>, sensor fusion, or control loop tuning</p></li><li><p>Knowledge of <strong>real-time systems</strong>, embedded software, or custom actuator control</p></li></ul><p></p><strong>Job Details</strong><ul><li><p><strong>Location:</strong> Cambridge, Ontario</p></li><li><p><strong>Work Environment:</strong> In-person (on-site at our Waterloo facility)</p></li><li><p><strong>Type:</strong> Full-time</p></li><li><p><strong>Compensation:</strong> Competitive salary (based on experience)</p></li><li><p><strong>Health Insurance:</strong> Provided</p></li><li><p><strong>Growth:</strong> Regular performance evaluations with potential for <strong>salary increases</strong> and <strong>stock option participation</strong></p></li></ul><p></p></p></p>