Senior/Staff Research Scientist, Physical AI - Dexterous Manipulation

Sanctuary AI

Vancouver

On-site

CAD 120,000 - 180,000

Full time

13 days ago

A complete application in a minute

Tailored resume and cover letter, ready to send.

Get past ATS filters

Job summary

Sanctuary AI is seeking an independent researcher to lead learning for a multi‑fingered, high‑DOF hand integrated with vision-language-action and world-model policies. You will design data collection, retarget demonstrations to the robot hand, and advance tactile sensing as a core input for learned behavior.

You will collaborate with hardware, controls, and teleoperation teams, deploy research on real hardware, and publish where appropriate, while exploring production-ready hand dexterity for

Qualifications

  • You are an independent researcher able to take a manipulation problem from question to working hardware result.
  • Hands-on experience with multi-fingered, high-DOF hands, tactile conditioned policies or dexterous teleoperation.
  • PhD in ML, robotics, or related field, or equivalent with first-author work deployed on physical systems.
  • Fluent in Python and PyTorch and comfortable inside a real robot software stack (ROS 2).

Skills

Independent researcher
Multi-fingered hand dexterity
Python
PyTorch
In-hand manipulation
Dexterous teleoperation

Education

PhD in ML/Robotics

Tools

ROS 2

Job description

Your New Role and Team

Sanctuary AI delivers production-ready Physical AI to industry: learned policies that run on commercially available robot arms and do real factory work today. Our edge is dexterity. We design and build a hydraulically actuated, high-degree-of-freedom hand with integrated tactile and proprioceptive sensing, and we train it to do the fine, contact-rich work that grippers cannot.

We are looking for a researcher to own how that hand learns. You will bring hand-level dexterity into vision-language-action and world-model policies, build the demonstration-data engine that feeds them, and make touch a first-class input to learned behavior. You will work alongside the hardware, controls, and teleoperation teams that build and run the hand, deploy on real customer tasks, and shape the next generation of the hand.

We hire globally for this role. We sponsor immigration, support relocation to Vancouver.

Why This Role?

A hand few researchers get to work on. Most robot-learning scientists spend their careers on parallel-jaw grippers or off-the-shelf research hands with no path to deployment. Here you train a human‑like, tactile‑sensed hand that our own team designs, builds, and iterates, and your results change what the next hand looks like.

The unsolved part of the stack. Foundation-model policies are only now reaching hand-level dexterity. Contact-rich, multi‑fingered manipulation is where the field’s hardest open problems and most visible results are, and it is the problem this role exists to solve.

Real work, not demos. Our policies run on industrial arms doing real tasks for real customers. You will see your research deployed, measured, and used.

Ownership. You set the dexterity learning agenda, with the hardware, controls, and teleoperation teams across the room and a modern GPU cluster behind you. You are the person who decides how the hand learns.

Publish. We encourage publication at top venues, support conference travel, and open-source where it makes sense.

A low‑friction move. We hire from anywhere, sponsor work permits through Canada’s fast‑track programs for tech talent, support permanent residency, and cover relocation. Vancouver is a safe, English‑speaking city on the Pacific, in the same time zone as the Bay Area, with public health care and strong schools.

Success Criteria
  • Dexterity in foundation-model policies. Bring high‑degree‑of‑freedom hand control into vision-language-action and world-model policies: action representation, conditioning on touch and proprioception, pre‑training and fine‑tuning, and on‑robot evaluation.
  • The dexterous data engine. Shape how demonstrations from human hands and teleoperation are collected, retargeted to the robot hand, filtered, and scaled into training data. Decide what data matters and how much of it we need.
  • Touch in the loop. Make tactile and proprioceptive sensing a first‑class input to learned policies, from representation and fusion through deployment on hardware.
  • Research leadership. Set the dexterity learning agenda, build rigorous hardware evaluation, publish where it makes sense, mentor other researchers, and feed what you learn back into the design of future hands. Simulation and reinforcement learning are tools in the kit where they help.
Your Experience
Qualifications
  • You are an independent researcher. You have taken a manipulation problem from question to working result on real hardware, with the method, experiments, and evaluation designed by you. You can scope and prioritize your own work from a high‑level goal, recognize when a direction is failing, and communicate results clearly to engineers who are not ML researchers.
  • You know the difference between a gripper and a hand. You have hands‑on experience with multi‑fingered, high‑degree‑of‑freedom hands, real or simulated: in‑hand manipulation, finger gaiting, tactile‑conditioned policies, or dexterous teleoperation and retargeting. You understand what changes in action‑space design, contact modeling, observation choices, and transfer when the end effector has twenty or more degrees of freedom.
  • You have the track record. A PhD in machine learning, robotics, or a related field, or equivalent experience, with first‑author work in robot learning and policies you have deployed on physical systems.
  • You have depth in at least two of:
  • You are fluent in Python and PyTorch and comfortable working inside a real robot software stack (ROS 2 or similar).
  • Strong pluses
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
Get your free, confidential resume review.

or drag and drop your file here.

Similar jobs

Similar jobs worth comparing

Senior/Staff Research Scientist, Physical AI - Dexterous Manipulation
Senior/Staff Research Scientist, Physical AI - Dexterous Manipulation

Sanctuary • Vancouver

On-site
CAD 150,000 - 230,000
Machine Learning Engineer
Machine Learning Engineer

Sanctuary • Vancouver

On-site
CAD 110,000 - 170,000
Principal Scientist, Physical AI
Principal Scientist, Physical AI

Sanctuary Corp • Vancouver

On-site
CAD 140,000 - 220,000
Machine Learning Research Intern (Reinforcement/Imitation Learning)
Machine Learning Research Intern (Reinforcement/Imitation Learning)

Sanctuary AI • Vancouver

On-site
CAD 20,000 - 27,000
Robotics Software Engineer, Teleoperation Toronto GTA · New York · San Francisco →
Robotics Software Engineer, Teleoperation Toronto GTA · New York · San Francisco →

Mecka • Toronto

On-site
CAD 120,000 - 160,000
Robotics Software Engineer, Teleoperation
Robotics Software Engineer, Teleoperation

mecka • Toronto

On-site
CAD 110,000 - 140,000
Robotics Engineer, Teleoperation
Robotics Engineer, Teleoperation

Mecka • Toronto

On-site
CAD 110,000 - 160,000
Machine Learning Research Intern (Reinforcement/Imitation Learning)
Machine Learning Research Intern (Reinforcement/Imitation Learning)

Sanctuary • Vancouver

On-site
CAD 20,000 - 31,000
Machine Learning Research Intern (Reinforcement/Imitation Learning)
Machine Learning Research Intern (Reinforcement/Imitation Learning)

Sanctuary AI • Vancouver

On-site
CAD 17,000 - 26,000
Senior Software Engineer, Real-Time Systems
Senior Software Engineer, Real-Time Systems

Sanctuary • Vancouver

On-site
CAD 120,000 - 170,000