OpenAI Returns to Robotics: Four Core Positions Open for Recruitment

OpenAI is making a high-stakes return to robotics, signaling its ambition to build physical AI systems that interact with the real world. The company has posted four core engineering positions—Electrical Engineer, Simulation Engineer, Actuator Design Engineer, and Control Systems Software Engineer—in a clear push to develop tangible robots. According to the job listings, base salaries for these roles range from $210,000 to $310,000 annually, with additional equity and options. This recruitment drive comes just weeks after OpenAI brought on Tairan He, a well-known robotics expert and tech influencer with over 500,000 followers, as part of its new Robotics team.

The move revives OpenAI’s earlier foray into robotics. Between 2017 and 2019, the company developed Dactyl, a five-fingered robotic hand trained using reinforcement learning and domain randomization. Dactyl could manipulate objects, flip a toy block, and even solve a Rubik’s cube single-handedly, establishing a benchmark for dexterous manipulation. However, OpenAI shut down its robotics team around 2020, citing the scarcity of robot training data compared to the vast textual and code data available online. The company redirected its focus to large language models, culminating in the ChatGPT series.

Now, with advances in simulation and AI, OpenAI is re-entering the field. The new team, led by Aditya Ramesh, has evolved from a world simulation research project into OpenAI Robotics. The current focus includes teleoperation for data collection, robot learning, and household manipulation tasks. Sam Altman has stated that AI should enter the physical world to assist humans, with short-term goals of building robots for skilled workers and long-term aspirations of making personal robots ubiquitous.

OpenAI faces fierce competition. Google DeepMind has consistently pursued robotic foundation models, Tesla is nearing production of its Optimus humanoid robot, and Figure AI—which secured nearly $1.7 billion in funding—recently completed extended continuous operation tests with zero failures. Against this backdrop, OpenAI is accelerating its efforts.

The robotics team includes several prominent Chinese researchers. Xingyu Lin, who joined in August 2024, holds a PhD from Carnegie Mellon University and contributed to the low-cost teleoperation framework GELLO and the HumanoidBench benchmark. Tairan He, a PhD from CMU, developed Omni H2O for whole-body humanoid control and also runs the popular tech podcast Whynot TV. Lawrence Yunliang Chen, a UC Berkeley PhD, focuses on robot learning and manipulation. In simulation and datasets, researchers like Chengshu Li (Stanford) and Hang Yin (Stanford Vision and Learning Lab) are building humanoid benchmarks and simulation environments for household tasks. For world model integration, PhDs such as Pengchuan Zhang (Tsinghua) and Jialiang Zhao (MIT CSAIL) are working on combining visual perception, world models, and robotics.

With this talent and renewed focus, OpenAI is signaling that the robotics race in Silicon Valley is entering a new, more intense phase.