Altman is recruiting. robot engineer In Silicon Valley.
Adopted by OpenAI last month He Tailana robotics technology expert and technology blogger with over 500,000 followers. now, OpenAI Robotics has published jobs for four core positions at once.
Electrical engineer, simulation environment engineer, actuator design engineer, control system software engineer –
We are preparing to build a real physical robot.
From the underlying circuit board to the entire control system, all link locations are open at the same time, making the layout intent clear at a glance.
The salary offer is also quite good. The base annual salary for some positions reaches between $210,000 and $310,000, which equates to up to more than RMB2.2 million, with additional stock options.
In fact, there were rumors in early May that OpenAI was planning to split its robotics and consumer hardware divisions. This large-scale adoption could open up new business opportunities.
However, the layout of the embodied intelligence track is not a hasty move. As of 2019, OpenAI had already launched a robotics project…
OpenAI reboots robotics
From 2017 to 2019, OpenAI focused on robotic hand projects. dactylIt used reinforcement learning to train a five-fingered bionic robotic hand to perform various object manipulation tasks.
For this project, we used the flexible hardware Shadow Hand. Utilizing reinforcement learning and automatic domain randomization techniques, we completed training on large data in a simulated environment and then transferred and used it on a real machine.
At that time, the robot hand not only achieved block flipping, but also overcame the problem of solving a Rubik’s cube with one hand. Able to operate stably even in the presence of external interference, it once became the industry benchmark for dexterous robot movement.
However, this project was later canceled.
Around 2020, OpenAI temporarily shut down its robotics team.
The reason is that the internet world has almost infinite text and code data, whereas the robot’s training data was scarce and the iteration speed was slow. In contrast, progress on the large-scale model route was faster and closer to the AGI goals that OpenAI was pursuing at the time.
Over the next few years, OpenAI was completely transformed into a language model, and the ChatGPT series of language models was a huge success.
However, in reality, Dactyl was also a global benchmark for reinforcement learning + dexterity at the time, and almost defined the mainstream technology route of “simulation training → actual implementation” in the robotics field.
This also inspired later ideas for “simulation-first” robot training at Google DeepMind, Tesla, and Figure AI.
Over time, competitors have already taken the lead.
Google Deep Mind We have always been engaged in research and development of basic robot models.
mass production of tesla optimus is approaching, and the production line is ready at the Fremont plant.
Diagram AIhas received nearly $1.7 billion in funding and recently completed ultra-long continuous operational testing with zero failures throughout the process.
In the face of the biggest AI entrance into the physical world, OpenAI can’t help but hit the ground running again.
A global simulation research project led by Aditya Ramesh has officially become OpenAI Robotics.
The team is currently in open recruitment mode. Four key engineering positions are completely open: electrical, simulation environment, actuator design, and control system software. We adhere to the R&D idea of thoroughly combining hardware R&D and machine learning, and are doing our best in the field of robotics.
Altman said artificial intelligence should enter the real world and assist humans in the physical world.
In the short term, we want to develop robots that provide services to technical workers and contribute to the construction of future infrastructure.
The long-term goal is for personal robots to be in every household and solve many of life’s trivial problems.
Chinese members join OpenAI’s robotics team
Over the past year, OpenAI has continued to expand its robotics lab and also plans to build a second experimental site.
Judging from the recruitment information and the team’s research direction, OpenAI’s robotics team is currently focused on remote control data collection, robot learning, and home scenario manipulation tasks.
In this team, Chinese researchers are central to three core research and development lines.
The first mainline focuses on robot learning and dexterity.
Hayashi Seyujoined the team in August 2024 and is an early core member of OpenAI’s robotics direction.
He graduated from Peking University with a major in computer science. After receiving his PhD from Carnegie Mellon University’s Robotics Institute, he completed his doctoral research at the Berkeley BAIR Institute, and has been deeply involved in the general field of robotics.
He was deeply involved in the development of GELLO, known as a low-cost remote control framework, and HumanoidBench, an evaluation benchmark for humanoid robots.
He TailanHe officially joined the team this spring and holds an undergraduate degree in computer science from Shanghai Jiao Tong University and a PhD from Carnegie Mellon University’s Robotics Institute.
The Omni H2O technology he developed explores the full-body coordination capabilities of humanoid robots, allowing them to complete movements and operations collaboratively under a unified framework.
In addition, He Tairan is also a technology blogger with over 500,000 followers and runs a famous technology podcast why not tv.
Lawrence Yunliang ChenHe holds a PhD from Berkeley and is from the AUTOLAB and BAIR laboratories, specializing in the direction of robot learning and manipulation.
He previously interned at NVIDIA and currently works at OpenAI on robot learning and operations research.
The second major line focuses on simulation, evaluation benchmarks, and data set construction.
Lee Seong-sooA graduate of Stanford University, he joined the team in June 2025 and has spent many years researching and developing evaluation benchmarks for humanoid robots suitable for housework scenarios.
Inhan From the Stanford Visual Learning Institute, he has previously participated in home activity simulation and embodied intelligence assessment projects such as BEHAVIOR and BEHAVIOR – 1K. His research directions at OpenAI include simulated environments, remote operations, and data collection.
The third major direction is to transfer world simulation technology to the robotics field.
Zhangpeng RiverA PhD in Mathematics from Tsinghua University, he worked at Meta FAIR for nearly four years and was deeply involved in the core research and development of major projects such as SAM and Llama.
Since joining OpenAI in February 2026, he has researched world simulation and robotics, exploring the integration of visual recognition, world models, and robotic systems.
Zhao JialiangThe MIT CSAIL Ph.D. has also publicly stated that he is joining OpenAI’s robotics lab and is focused on research and development of general agents for the real physical world.
In other words, OpenAI Robotics, which already has many big names, is still recruiting –
Competition is heating up once again in Silicon Valley’s robotics arena.
Reference links: [1]https://x.com/sama/status/2061117302528188712[2]https://cryptobriefing.com/openai-robotics-hiring-engineers/
This article is from the WeChat official account “QbitAI”. Author: Focuses on cutting-edge technology. Republished with permission by 36Kr.
