Dexmal
Posted 4mo ago

具身端到端算法工程师

Dexmal
Beijing, Beijing, China
OnsiteFull Time
Responsibilities
  • designing frameworks
  • implementing strategies
  • detecting failures
Requirements
  • Bachelor's in robotics/automation/CS/AI,3+ years robot algorithm or motion planning experience
  • Strong math, optimization and control theory
  • Python and C++
  • Experience with VLA/Diffusion Policy and robot execution frameworks
Technical tools mentioned
PythonC++VLADiffusion PolicyBehavior TreeState MachineTAMP

Job description

负责机器人端到端执行层的算法设计与工程落地,在 VLA等大模型输出的高层指令与机器人底层运动执行之间构建可靠的「执行桥梁」。通过融合规则、传统算法与学习策略,保障机器人在真实场景中的任务执行稳定性与安全性。
1. 负责机器人端到端任务执行框架的设计与开发,承接大模型输出的高层动作指令并转化为可靠的底层执行
2. 设计并实现基于规则与算法的符合策略,在模型置信度不足或场景超出模型泛化范围时提供安全兜底
3. 开发任务执行的异常检测与恢复机制,识别执行失败、环境突变等异常状态并触发相应处理逻辑
4. 针对具体落地场景(如工厂搬运、装配等),设计场景专属的混合执行策略,结合感知反馈实现自适应调整
5. 打通执行算法与运动控制模块的接口,保障算法策略可落地到真实硬件
7. 分析进厂测试中的执行失败案例,持续迭代符合策略与兜底规则,提升任务完成率

基本要求
1. 本科及以上学历,机器人、自动化、计算机、人工智能相关专业
2. 3 年以上机器人算法或运动规划开发经验
3. 熟悉机器人任务规划与执行框架(行为树、状态机、TAMP 等)
4. 具备扎实的数学基础,熟悉优化方法、概率模型与控制理论
5. 熟练使用 Python 与 C++,有算法工程化落地经验
6. 有 VLA、Diffusion Policy 等端到端模型与传统控制混合部署的实际经验
7. 熟悉机械臂操纵、灵巧手抓取等具体任务场景的执行策略设计
8. 有真实工业场景落地经验优先

About Dexmal

Develops end-to-end embodied AI software and robotics hardware.

Year founded
2025
Employees
150
Organization type
Private
Latest investment
Series A (2025)
Headquarters
CN

Similar jobs

End-to-End Algorithm Engineer roles near Beijing, Beijing
3d
Save
Mark Applied
Hide
智能辅助驾驶-一段式端到端算法(问题迭代方向)
Beijing or Shanghai
OnsiteFull Time
NIO
NIONYSE: NIO: Designs and manufactures smart, premium electric vehicles.
3+ YOEMaster's or doctoral degree in robotics, automation, computer science, or computer engineering; 3+ years of intelligent driving algorithm development; experience with deep learning, data mining, video analysis, databases, C++, Python, and Lua.
CNN, Transformer, LLM, VLM, SFT, DPO, GRPO, C++, Python, Lua
3w
Save
Mark Applied
Hide
端到端算法工程师
Beijing or Shenzhen or Shanghai
OnsiteFull Time
Momenta
MomentaHKEX: Momenta: An artificial intelligence developing autonomous driving technologies and end-to-end driving models.
Master's degree or higher in computer science, software engineering, robotics, automation, or related fields; expertise in machine, deep, and reinforcement learning; proficiency in C++, Python, and PyTorch.
C++, Python, PyTorch
2mo
Save
Mark Applied
Hide
顶尖应届-端到端算法工程师-自动驾驶
Beijing, Beijing, China
OnsiteFull Time
Xiaomi
XiaomiHong Kong Stock Exchange: 1810: Designs and manufactures smartphones, consumer electronics, and electric vehicles.
Master's degree in CS/Electronic Information/Automation or higher, strong deep learning and computer vision foundation, PyTorch/TensorFlow experience, C++ and deployment skills, familiarity with multimodal models and model optimization.
PyTorch, TensorFlow, HuggingFace, OpenCV, C++, ONNX Runtime
7mo
Save
Mark Applied
Hide
泊车端到端算法工程师(L4)
Beijing or Shanghai or Suzhou
OnsiteFull Time
QCraft
QCraft: Develops autonomous driving solutions for robotaxis and vehicles.
Master's degree or above (PhD preferred). Strong in Python and C++, experienced with PyTorch, CNNs and Transformers, ML/CV knowledge, algorithm development and model deployment.
Python, C++, PyTorch, CNN, Transformer
1y
Save
Mark Applied
Hide
自动驾驶端到端算法工程师
Beijing or Hangzhou or Shanghai
OnsiteFull Time
PhiGent Robotics
PhiGent Robotics: Develops autonomous driving perception and intelligent software solutions.
Strong deep learning and computer vision background, excellent programming, familiarity with PyTorch, experience in autonomous driving and 3D perception, able to solve complex engineering problems.
PyTorch
3d
Save
Mark Applied
Hide
【27届校招】一段式端到端模型算法工程师
Beijing or Shanghai or Suzhou
OnsiteFull Time
QCraft
QCraft: Develops autonomous driving software and intelligent mobility solutions.
Master's or doctoral degree in computer science, electronic engineering, automation, mathematics, or related fields; strong Python algorithm development and PyTorch skills; top CV/AI publications or competition awards preferred.
Python, PyTorch