人形机器人造梦师 / Humanoid Dream Architect（仿真与合成数据）

岗位使命

为人形机器人构建“梦境”：用 Vibe Coding 和 Agentic Workflow 构建 GTA 级别的高保真仿真城市与合成数据管线，让机器人模型在梦中持续学习、快速成长。

你将负责

  • 用 Agent 构建城市级高保真仿真环境——建筑、街道、光照、天气、行人与动态交互元素，达到 GTA 级的规模与真实感，且完全可物理交互，让人形机器人能够在其中真实执行任务。
  • 设计并构建高效的合成数据 Agentic Workflow，让场景生成、数据采集、筛选与评估全自动运转。
  • 建设数据质量与真实感评估体系，确保合成数据真正提升模型能力。
  • 与 VLA 和运控团队协作，定义数据规格与接口，打通从仿真到真机的数据链路。

我们希望你

  • 有极强的 Agent Engineering / Vibe Coding 能力，善于用 Agent 构建高效的 Agentic Workflow，把复杂流程做成稳定可靠的自动化系统。
  • 熟悉 Isaac Sim/Lab、MuJoCo 等物理仿真工具链者优先，能构建完全可物理交互的机器人仿真场景尤佳。
  • 熟悉 3DGS（3D Gaussian Splatting）等场景重建与渲染技术者优先。
  • 有游戏 3D 建模、技术美术（TA）、程序化生成（PCG / Houdini）或游戏引擎（Unity/Unreal）开发经验者优先。
  • 有仿真环境、数字资产管线或大规模场景生成的实际构建经验。
  • 理解机器人模型训练的数据需求，能与模型团队高效协作。

---

Humanoid Dream Architect (Simulation & Synthetic Data)

Mission

Build dreams for humanoid robots: use vibe coding and agentic workflows to build GTA-scale, high-fidelity simulated cities and synthetic data pipelines — so robot models keep learning and growing in their dreams.

What You'll Do

  • Use agents to build city-scale, high-fidelity simulation environments — buildings, streets, lighting, weather, pedestrians, and dynamic interactive elements — at GTA-level scale and realism, fully physically interactive so humanoid robots can actually operate in them.
  • Design and build efficient agentic workflows for synthetic data, fully automating scene generation, data collection, filtering, and assessment.
  • Build data quality and realism assessment systems to ensure synthetic data genuinely improves model capability.
  • Work with the VLA and control teams to define data specs and interfaces, connecting the pipeline from simulation to real robots.

What We're Looking For

  • Exceptional agent engineering / vibe coding ability — skilled at building efficient agentic workflows, turning complex pipelines into stable, reliable automated systems with agents.
  • Familiarity with physics simulation toolchains such as Isaac Sim/Lab or MuJoCo is a strong plus — especially experience building fully physically interactive simulation scenes for robots.
  • Familiarity with 3DGS (3D Gaussian Splatting) or related scene reconstruction and rendering techniques is a plus.
  • Experience in game 3D modeling, technical art (TA), procedural generation (PCG / Houdini), or game engine development (Unity/Unreal) is a plus.
  • Hands-on experience building simulation environments, digital asset pipelines, or large-scale scene generation.
  • Understanding of the data needs of robot model training; able to collaborate effectively with model teams.