1,827,878open jobs
106,403companies
263,677added this week
Browse all
Salary
≈ $57k – $137k per year (Estimated)
Location
In office (Shanghai)
Seniority
Staff · 5+ years exp
Employment
Full-Time

Confirmed on the employer's own hiring board on Oct 19, 2026. First seen by Alion on Mar 12, 2026. AgiBot scores A on the Alion truth index.

Overview
Company
Impact
Profile match
Headquartered in Shanghai, China, AgiBot - also known as Zhiyuan Robotics - is an innovative artificial intelligence and robotics company specializing in general-purpose humanoid robots and embodied AI systems. The firm develops advanced hardware platforms and specialized operating software that integrate computer vision, motion control, and deep learning capabilities. By producing both bipedal and wheeled robotic systems, it provides scalable automation solutions designed for industrial manufacturing, commercial environments, and research applications.

岗位职责 / Responsibilities

职位背景

我们正在建设一个面向 具身智能(Embodied AI)与机器人学习 的大规模研究中心,拥有 大规模 GPU 集群与真实机器人数据闭环。团队致力于研发 多模态大模型(Multimodal Foundation Model)、VLA(Vision-Language-Action)、强化学习(Reinforcement Learning)与世界模型(World Model) 等核心技术。

我们在招聘 ML Infrastructure Lead,负责设计和建设支持大规模模型训练的 机器学习平台与分布式训练基础设施,包括 训练系统、GPU 集群调度、实验平台与数据训练 pipeline,以支持研究团队在大规模计算环境下高效进行模型训练与算法迭代。

工作职责

负责机器学习基础设施与训练平台的架构设计与建设

设计并维护 大规模 GPU 集群训练系统(H100/H200 等)

构建支持 大模型(LLM / Multimodal / VLA) 的分布式训练系统

搭建 训练 pipeline(dataset → preprocessing → training → evaluation → deployment)

设计和实现 实验管理系统(Experiment Tracking)、模型版本管理(Model Registry)

优化 GPU 利用率、通信效率与训练稳定性

与研究团队合作支持 大模型训练、强化学习训练与多模态模型训练

与机器人系统团队合作,实现模型 推理部署(Inference / Serving)

组建并领导 ML Infrastructure 团队

任职要求 / Requirements

任职要求

计算机、人工智能或相关专业 硕士及以上学历

5年以上机器学习工程或 AI 平台经验

熟悉主流深度学习框架:PyTorch / JAX / TensorFlow

熟悉分布式训练技术:FSDP / Megatron / DeepSpeed / Horovod

熟悉 GPU 集群与训练调度系统:Slurm / Kubernetes / Ray

熟悉机器学习训练 pipeline 与数据系统:dataset pipeline / experiment tracking / model registry

熟悉 GPU 计算优化:CUDA / NCCL / mixed precision / memory optimization

具备良好的系统架构能力与团队协作能力

加分项

有大模型训练(LLM / Multimodal Foundation Model) 经验

有强化学习(RL / RLHF / Offline RL) 训练平台经验

有自动驾驶 ML Infra 或 AI 平台团队经验

有管理 数百至上千 GPU 集群训练任务 的经验

有构建 AI Training Platform / ML Platform 的经验

Free account
Stop reading job ads. Get the ones that fit.
One free account turns this page into a shortlist built around your stack, your level and your pay.
Match on every job. Stack, seniority, pay and location, scored against your profile.
1,827,878 open roles. Read straight off company career pages, refreshed every day.
Unlimited applications. Every one you send is tracked in one place, on-site or on a company board.
3 tailored CVs a month. Rewritten for the exact job you are applying to. Included free.
Create a free account Continue with Google
Free forever. No card. Under a minute.

Your match

How well do you fit this role?
Two answers are enough for a real match. No account needed.
Check my fit
Answers stay in this browser until you create an account.

Recommended for you based on this role

AI/ML
Similar stack
Same company
Shanghai
≈ $48k – $117k per year (Estimated) • In office • 5+ years exp • Beijing
AI/ML
Fine-tuning
Quantization
Computer Vision
Apply
In office • Internship • Beijing
AI/ML
LLM
Apply
In office • Internship • Beijing
Go
Java
C++
AI/ML
Cursor
Claude
LLM
RAG
Apply
In office • Full-Time • Beijing
C++
AI/ML
NLP
Apply
In office • Internship • Chengdu
Python
Java
PHP
C++
Apply
In office • Shanghai
AI/ML
LangGraph
LangChain
DeepSpeed
Model Context Protocol
vLLM
Function Calling
SGLang
Transformers
RAG
Mixture of Experts
DPO
SFT
PPO
GRPO
FSDP
KV Cache
Apply
In office • Full-Time • Shanghai
Python
C++
C++
PyTorch C++
AI/ML
PyTorch
Vision-Language-Action
World Models
Apply
JAVA后端工程师 6 hours ago
In office • Full-Time • 2+ years exp • Shanghai
QA
TestNG
Apply
≈ $30k – $70k per year (Estimated) • In office • Full-Time • 3+ years exp • Master's Degree • Shanghai
Apply
≈ $16k – $30k per year (Estimated) • In office • Internship • Shanghai
Apply
See all jobs
This is one of many
1,827,878 more open roles from verified company boards, updated every day.