Senior Research Scientist, Robot Learning

present of AI; San Francisco (CA); Hybrid; USD 180,000 - 240,000; Full time

Job summary

Hub is seeking a Senior Research Scientist in Robot Learning, based in San Francisco or Toulouse. The role blends hands-on coding, robot trials, and experimental design to advance policy learning with real robots.

A PhD and several years of research experience with publications in robot learning are expected, including experience training and evaluating policies on physical robots. The ideal candidate will turn failures into testable questions, mentor junior researchers, and help shape the lab's

Qualifications

• This role requires turning real robot failures into testable questions.
• Candidates should have published in robot learning and trained/evaluated policies on physical robots.

Responsibilities

• Turn real robot failures into testable questions and answer with careful experiments.
• Mentor others and help shape the lab's research agenda.
• Balance hands-on coding with robot trials to advance policy learning.

Skills

Robot learning

Policy testing on real robots

Hands-on coding

Mentoring

Experiment design

Education

PhD in robotics / computer science

Tools

Python

ROS

Job description

Senior Research Scientist, Robot Learning

AI & Data · San Francisco, CA, US / Toulouse, France · Full-time · Senior level

This role asks for a researcher who can turn real robot failures into testable questions and answer them with careful experiments. A strong candidate holds a PhD plus several years of research, has published in robot learning, and has trained and evaluated policies on physical robots rather than only in simulation. The work mixes hands-on coding and robot trials with shaping the lab's research agenda and mentoring others.

How it fits the AI era

Robot learning is core to the AI era, and this role sits where data, models and real machines meet. AI speeds up experiment loops and policy training, but choosing which failures matter, designing fair tests and judging results stay human. The job keeps room to think, publish and open source, so judgment and rigour matter more than raw output.

How it builds AI experience

The candidate would gain experience turning large egocentric video and teleoperation data into working robot behaviour, including imitation learning, vision language action models and world models. Building this means showing policies tested on real robots and work on dataset or benchmark design. Experience with humanoids or complex embodied systems and open source contributions would strengthen a case.

Worth showing

• Real robot policy training


• Robot learning publications


• Imitation learning and VLAs

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 Hub

Global physical crowdsourcing network capturing verified egocentric multimodal video datasets for robotics and frontier world models, per the company.