【AI Infra】数据平台工程师(J12899)

免费注册后解锁"你的匹配分 vs 高分画像"与个人技能缺口。

1、做什么：
建设具身智能专用的数据闭环基础设施，覆盖"采集→传输→处理→存储→标注→评测→回流"全链路，支撑多模态时序数据的高效流转与价值提炼。

2、核心职责：
- 端云一体数据闭环架构设计与搭建，支持断点续传、弱网同步
- 多模态时序数据处理引擎（清洗、微秒级时序对齐、质量评估）
- 数据血缘追踪与版本管理（数据集版本管理、模型-数据版本绑定、A/B 对比与回滚）
- 数据质量评价体系与可视化调试工具建设
- 计算机/数据工程 等相关专业，硕士/博士学历
- 熟悉分布式数据处理框架（Spark/Flink/Ray），有流批一体基础
- 了解消息队列（Kafka/Pulsar）、时序数据库、对象存储
- 有 ROS/ROS2 或机器人/自动驾驶数据经验者优先
- 对数据治理、版本管理有实际理解

加分项：
- 具身智能/机器人多模态数据处理经验
- AI 辅助标注、主动学习、自动数据挖掘经验
- 开源数据平台（Airflow/Dagster/Metaflow）深度使用

1. What to do:
Build data closed-loop infrastructure dedicated to embodied intelligence, covering the full pipeline of "collection → transmission → processing → storage → annotation → evaluation → feedback" to support efficient flow and value extraction of multimodal time-series data.

2. Core responsibilities:
- Design and build end-cloud integrated data closed-loop architecture, supporting resumable transfers and weak-network synchronization
- Multimodal time-series data processing engine (cleaning, microsecond-level time-series alignment, quality assessment)
- Data lineage tracking and version management (dataset version management, model-data version binding, A/B comparison and rollback)
- Build data quality evaluation systems and visual debugging tools
- Master's/PhD degree in Computer Science/Data Engineering or related fields
- Familiar with distributed data processing frameworks (Spark/Flink/Ray), with a foundation in stream-batch unification
- Understanding of message queues (Kafka/Pulsar), time-series databases, and object storage
- Experience with ROS/ROS2 or robotics/autonomous driving data is preferred
- Practical understanding of data governance and version management

Bonus points:
- Experience in embodied intelligence/robotics multimodal data processing
- Experience in AI-assisted annotation, active learning, and automated data mining
- Deep use of open-source data platforms (Airflow/Dagster/Metaflow)

⚑ 岗位信息收录自公开招聘页（国内官方招聘渠道），以雇主官网为准