Research Scientist in Robotics Foundation Models at Ai2, the Allen Institute for AI (Seattle, USA)
Job Description
Ai2, the Allen Institute for AI, is hiring a research scientist in robotics to study foundation models for robots, using large-scale simulation and real robots, and to release the methods, datasets, benchmarks and models openly.
At a glance
- Position: Research Scientist, Robotics
- Where: Ai2 offices, Seattle, Washington, USA (on site)
- Pay: Base salary US$167,030–260,570 a year, plus bonus
- Type: Full time
- Employer type: Non-profit AI research institute
- Apply: open until filled (re-checked 13 November 2026)
About the robotics research scientist role at Ai2
- Investigate core research problems in foundation models for robotics, with fully open methods, datasets, benchmarks and models
- Experiment with large-scale simulation tools and with robots in the real world
- Work on robot manipulation, large-scale training in simulation, sim-to-real transfer, robot foundation models and skill learning
- Lead projects, mentor interns and pre-doctoral researchers, and write and present papers
- Collaborate with other Ai2 teams in computer vision and language models, and with universities
The posting says Ai2 places no restrictions on publication, and the robotics team publishes at CoRL, ICRA, RSS, CVPR, ICCV, ECCV, NeurIPS, ICLR and ICML. Ai2’s Seattle office is close to the University of Washington, and the posting welcomes applicants from outside the United States.
Why this role matters
Robot foundation models are one of the most contested frontiers in AI, and most of the work happens inside companies that release little. Ai2 commits to open models and data, so a scientist here can publish freely while working at a scale few university labs can match. That combination is rare, and it makes this post a strong option for a robotics PhD choosing between academia and industry.
What Ai2 is looking for
- A strong foundation, typically at PhD level, in robot manipulation, planning and control, reinforcement or imitation learning, real-world robot systems, simulation and sim-to-real transfer, or world and foundation models
- A strong publication record in AI-related venues such as CoRL, ICRA, RSS, CVPR, ICCV, ECCV, NeurIPS or ICLR
- Strong software engineering skills, with experience of real robot systems or deep-learning frameworks
Nice to have
- Research on large-scale multi-node training, training dynamics, efficiency, data curation, multimodal debugging or post-training
- Contributions to open-source research libraries
- Community service such as organising workshops or tutorials
Pay and location
The posting gives a base salary range of US$167,030 to US$260,570 a year, plus bonus plans. The role is based in Seattle and the posting says scientists are expected to work from the office there. Ai2 participates in E-Verify for US work authorisation.
How to apply
Apply through the official Ai2 job posting. Ai2’s posting does not give a closing date, so the role is open until filled; ResearchJobs.in will re-check this listing on 13 November 2026. Check the posting for location and work-authorisation details before you apply.
See all our artificial intelligence jobs, or browse more research jobs on ResearchJobs.in.
Hiring institution: Allen Institute for AI (Ai2)
Official advertisement: job-boards.greenhouse.io
How to prepare for this application
- Open work counts: point to code, datasets or models you have released, since open research is central to Ai2's mission.
- Sim-to-real in depth: be ready to discuss what broke when a policy moved from simulation to hardware, and how you fixed it.
- Scaling questions: prepare a view on how data diversity and model size affect generalisation in manipulation.
- Know recent Ai2 robotics papers: read the team's recent releases so you can propose a project that builds on them.
- Research talk: expect a job talk; keep it to one clear thread with honest failure cases.
About Ai2
The Allen Institute for AI (Ai2) is a non-profit AI research institute in Seattle, founded in 2014 by Microsoft co-founder Paul Allen. It is known for releasing fully open AI models, data and tools, including the OLMo language models and the Molmo multimodal models, and it works on language models, computer vision, robotics, scientific literature tools and conservation technology.
