ByteDance
Posted 3mo ago

大模型数据工程师-Code/Agent

ByteDance
Beijing, Beijing, China
OnsiteFull Time
Responsibilities
  • building data
  • supporting platform
  • designing architecture
Requirements
  • Design and build large-model data production
  • Agent workflows
  • Automated labeling and evaluation, and platform systems
  • Strong engineering experience with Go/Python/TypeScript and large-model data pipelines
Technical tools mentioned
GoPythonTypeScript

Job description

1、负责大模型数据生产体系建设及Agent化能力落地,覆盖Code、Agent、Application等重点方向,从需求理解、数据合成、环境构建、质量评估到训练反馈回流,支撑模型在代码、多模态、工具调用、GUI操作和应用任务上的效果迭代;建设面向GeneralAgent、GUI Agent、Tool Use、Workspace/Env Scaling等场景的Workflow编排、任务调度、轨迹分析、自动评测和自迭代系统,提升复杂任务数据的规模、质量和可复用性;
2、参与Code数据生产平台与业务产线建设,支撑VLM Coding、Frontend、SFT/MidTrain、DataScaling、Dogfooding/User feedback、Stone等方向,将真实用户回流、环境重建、长程任务和多模态代码任务转化为可训练、可验证、可持续迭代的数据资产;推动Application数据生产能力建设,面向豆包办公、群聊、企业问答等应用场景,抽象通用数据生产流程、标注工艺、质量门禁和评测回流机制,提升应用场景数据生产的标准化和自动化水平;
3、推动数据平台智能化演进,建设AI辅助的智能标注、数据Mock、质量校验、价值评估、错误修正和任务分发能力,降低人工链路成本,提升数据生产稳定性和规模化效率;
4、负责复杂中后台系统的架构设计与渐进式治理,在引入大模型辅助研发和Agent自动化能力的同时,建立清晰的业务边界、前后端契约、测试验证、权限隔离和风险防御机制;
5、跟踪大模型、Agent、代码生成、多模态、自动化评测等前沿方向,与算法、产品、评测、业务团队协作,将技术探索转化为真实业务产线和模型效果收益。

1、本科及以上学历,计算机、人工智能、软件工程等相关专业,具备大模型研发、数据工程、Agent系统、复杂中后台或平台型系统研发经验;
2、具备扎实工程能力,熟悉Go/Python/TypeScript中至少一种,能够独立完成复杂业务系统设计、状态流转建模、高复用组件抽象、性能与稳定性治理;
3、理解大模型数据生产链路,熟悉数据合成、标注平台、训练数据管道、质量评估、数据回流、Benchmark或自动化评测中的一种或多种,有真实业务落地经验优先;具备Agent/Workflow/Tool Use/GUI自动化/Sandbox/Runtime相关经验者优先;熟悉Prompt优化、轨迹分析、记忆系统、工具协议、任务编排或自迭代机制者优先;具备AI Coding、VLM Coding、长程任务、多模态数据、真实用户反馈数据、应用数据生产经验者优先;
4、具备强烈的契约优先和边界防范意识,能在使用AI工具提升研发效率的同时,通过模块边界、自动化测试、质量门禁、灰度发布和可观测体系保障系统可靠性;
5、具备良好的跨团队协作能力和自驱力,能从业务问题出发定义技术方案,推动算法、产品、评测和工程团队形成可落地的生产闭环。

About ByteDance

Global technology specializing in AI-powered content platforms.

Similar jobs

Large Model Data Engineer roles near Beijing, Beijing
6mo
Save
Mark Applied
Hide
大模型数据工程师
Beijing or Shanghai
OnsiteFull Time
智元机器人
智元机器人: Developing embodied AI and intelligent humanoid robotics solutions.
5+ YOEMaster's or higher in CS/AI/Robotics,5+ years large-scale data engineering or large-model fine-tuning,expertise in VLMs,2D/3D vision and point cloud processing,distributed systems and online labeling.
VLM, CLIP, LLaVA, SFT, RLHF
1y
Save
Mark Applied
Hide
百川智能-大模型数据工程师
Beijing, Beijing, China
OnsiteFull Time
Baichuan AI
Baichuan AI: Private Chinese AI developing large language models for enterprises and the public.
2+ YOEBachelor's degree and 2+ years in web crawling or big data; skilled in Scrapy, Hadoop/Spark/Hive/HBase, Python/Java/Go, and ML frameworks; able to build and optimize large-scale data pipelines.
Fiddler, Scrapy, Hadoop, Spark, Hive, HBase, Python, Java, Go, PyTorch, TensorFlow
1w
Save
Mark Applied
Hide
大模型数据开发工程师
Beijing, Beijing, China
OnsiteFull Time
ModelBest
ModelBest: AI startup specializing in on-device foundation models.
3+ YOEBachelor's degree or higher in a computer-related field, 3+ years of data development experience, and proficiency in Python and distributed frameworks such as Spark and Ray.
Python, Apache Spark, Ray
2mo
Save
Mark Applied
Hide
顶尖应届-具身大模型数据工程师-机器人事业部
Beijing, Beijing, China
OnsiteFull Time
Xiaomi
XiaomiHong Kong Stock Exchange: 1810: Global technology specializing in consumer electronics and smart manufacturing.
Expertise in large-scale video processing, action recognition, CLIP/VLM models, Python, PyTorch/TensorFlow, and distributed frameworks (Ray, Spark); prompt engineering or model distillation experience preferred.
Python, PyTorch, TensorFlow, Ray, Spark, CLIP, VLM
1w
Save
Mark Applied
Hide
大模型预训练数据算法工程师
Beijing or Shanghai
OnsiteFull Time
MiniMax
MiniMaxHKEX: 0100: Global AI developing multimodal foundation models and applications.
Strong Python programming, data algorithm or large-scale data engineering experience, large-scale data processing, data quality analysis, and basic machine learning and large language model pretraining knowledge.
Python