Staff Research Scientist, Dexterous Manipulation at Optalis

Vancouver, British Columbia, Canada -

Full Time

Start Date

Immediate

Expiry Date

09 Aug, 26

Salary

0.0

Posted On

11 May, 26

Experience

5 year(s) or above

Remote Job

Yes

Telecommute

Yes

Sponsor Visa

Skills

Python 3.8, PyTorch, TensorFlow, ROS2, Reinforcement Learning, Imitation Learning, Sim-to-real Transfer, Continual Learning, Parallel Simulations, Robotic Manipulation, Jira, Confluence, GitLab, Haptic Sensing, Proprioceptive Sensing, Machine Learning

Industry

Description

Your New Role and Team Sanctuary, a world leader in building AI-based control systems for humanoid robots, is seeking a Staff Research Scientist to join our team in engineering and innovating unique robotic manipulation tasks. As a Staff Research Scientist, your role will involve choosing the most cutting-edge methods, creating training and data collection systems, overseeing the evaluation of these algorithms in simulated environments, and implementing them on our robots in real-world situations. You will also enjoy the exclusive chance to make a meaningful impact by working with novel haptic and proprioceptive sensing techniques, thanks to our in-house robot with dexterous hands. Success Criteria * Create, develop, and enhance cutting-edge Reinforcement Learning (RL) and Imitation Learning (IL) algorithms and evaluate their performance in practical applications * Stay current with the latest developments in RL/IL techniques and their application in robotics * Identify, communicate, and lead research initiatives that show promise to the wider ML team * Discover strategies for enhancing current RL/IL learning processes, considering key performance metrics like sample efficiency, speed, computational resources, and scalability * Devise RL/IL training and data collection pipelines to expedite implementation on physical robots * Collaborate within a diverse team to devise innovative algorithms and investigate the root causes of errors in existing implementations Your Experience Qualifications * Ph.D. in Machine Learning, Computer Science, Applied Mathematics, or equivalent practical background in Reinforcement Learning and/or Imitation Learning * 5+ years of hands-on experience implementing and deploying robotic manipulation tasks, both in simulation and on physical robots * 5+ years of practical experience applying various Reinforcement Learning and/or Imitation Learning methods, with focus on robotics in the real world * 4+ years experience in developing and optimizing large-batch parallel simulations for Reinforcement Learning * Proven expertise in continual learning, employing adaptive model training to improve long-term performance and accuracy * Proven expertise in sim-to-real transfer * Experience in transitioning Machine Learning research and trained models into real-world production * Active involvement in integrating Machine Learning models into a robotics platform * A track record of publishing research in esteemed AI conferences such as ICRA, IROS and CORL Skills * Development with Python 3.8 or later * Working knowledge of PyTorch and/or TensorFlow * Familiarity with ROS2 * Expertise in use of Reinforcement Learning principles and their application * Experience with Atlassian tools; Jira, Confluence, or equivalent i.e. GitLab Traits * Above all else, a consistently positive attitude and a willingness to do whatever it takes to create robust solutions to complex problems * Strong leadership skills in organizing R&D work for ML projects * Eager to take on new challenges with tenacity and positivity * Patience, persistence, and attention to detail when resolving performance issues * Enthusiasm for bringing human-like intelligence to machines * Ability to drive development of new functionalities from concept to production * Ability to multitask and prioritize in a fast paced environment

Responsibilities

Develop and enhance cutting-edge Reinforcement Learning and Imitation Learning algorithms for humanoid robotic manipulation. Lead research initiatives and create data collection pipelines to transition models from simulation to real-world physical robots.