Build the next
embodied intelligence.一起构建下一代
具身智能。
We are building a Humanoid Foundation Model — connecting intelligence and embodiment through real robots, synthetic data and ambitious engineering.我们正在构建通用人形机器人基座模型，以真实机器人、合成数据和大胆工程，连接智能与具身。
Open roles招聘岗位
Full-time · Beijing/Shanghai (Humanoid Control: Beijing only)全职 · 北京/上海（运控岗位仅限北京）
RSI ResearcherRSI 研究员
Design self-improving research systems that help agents form hypotheses, run experiments and learn from results.构建能提出假设、执行实验并从结果中学习的自我改进研究系统，加速基座模型研发。
- Agents · evaluation · experiment design
- Strong coding and rapid prototyping
Agent Infra Full-Stack EngineerAgent Infra 全栈工程师
Build the harness, tools and product surfaces that let agents reliably execute work on our development machines.建设 Agent Harness、工具和产品体验，让 Agent 能在开发机上可靠执行任务并交付结果。
- TypeScript / React / Python / Go / Rust
- Distributed systems, observability and product craft
VLA Full-stack ResearcherVLA 全栈研究员
Train and deploy vision-language-action models for navigation, dexterous manipulation and loco-manipulation.负责 VLA 数据、训练与评测，将模型部署到真实人形机器人，覆盖导航、灵巧手和移动操作。
- VLM / VLA / multimodal modeling
- Pre-training, post-training or RL experience
AI Infra Full-Stack EngineerAI Infra 全栈工程师 · 训练与推理
Make training faster, serving lower-latency and the path from experiment to robot shorter.建设高性能、高可靠的训练与推理基础设施，缩短从实验到真机验证的周期。
- PyTorch · CUDA · Triton · NCCL
- Distributed training, serving and performance profiling
Humanoid Control Engineer / Researcher人形机器人运控工程师 / 研究员
Build the robot’s “motor cortex”: reusable full-body skills that turn intent into stable motion on hardware.构建人形机器人的“运动小脑”：把任务意图转化为稳定、连续、可迁移的全身动作。
- RL · whole-body control · sim-to-real
- C++ / Python / real-time robot systems
Humanoid Dream Architect人形机器人造梦师
Build high-fidelity simulation and agentic synthetic-data workflows, so robot models keep learning in their dreams.建设高保真仿真环境与合成数据 Agentic Workflow，让机器人模型在“梦境”中持续学习成长。
- Unity/Unreal · PCG / Houdini · 3D modeling
- Agentic workflow and data quality
Who we are looking for我们共同寻找的人
We care about what you have built, the hard problems you have solved and how quickly you turn ideas into working systems. Non-traditional backgrounds are welcome.我们更关注你做过什么、解决过什么困难问题，以及能否快速把想法变成真实运行的系统。我们欢迎非传统教育背景。
- Curious about intelligence and robots真正热爱智能、机器人和工程创造
- Own outcomes end to end主动拆解目标并对结果负责
- Ship, measure, learn and iterate快速执行、验证、复盘和迭代
- 1–3 years experience welcome欢迎毕业 1–3 年的候选人