[Skip to main content](#main-content)

[OpenNash Jobs](https://jobs.opennash.com/)
[Hiring report](/reports/hiring/) · [All Kuaishou Technology jobs](https://jobs.opennash.com/companies/kuaishou-technology/) · [Search jobs](/search/)

Kuaishou Technology · Beijing

# 多模态大模型后训练算法专家​

Beijing 

Listed on Kuaishou Technology’s official careers site. Listed as of October 19, 2026. Check that it is still open and apply there.
[Apply on Kuaishou Technology’s careers site →](https://zhaopin.kuaishou.cn/#/official/social/job-info/26279)

[OpenNash Jobs](https://jobs.opennash.com/) / [Kuaishou Technology](https://jobs.opennash.com/companies/kuaishou-technology/) / 多模态大模型后训练算法专家​

## Job description
- 主导多模态大模型（Keye）基座模型的后训练全流程优化，包括指令微调（SFT）、强化学习（RLHF/DPO/PPO）、模型蒸馏等技术的研发与工程实现，持续提升模型在通用领域及垂直领域（如短视频理解、电商理解）的泛化能力和安全性；
- 参与多模态合成数据生成（覆盖纯文本、图文及视频等内容理解维度）、建立数据质量评估体系，设计数据蒸馏流程，推动​​数据-模型-评测​​迭代优化；
- 面向短视频社区、电商内容理解等场景，解决后训练阶段的核心挑战，​​如​​跨模态对齐一致性、长上下文推理、多任务指令冲突优化以及多模态智能推理探索等；推动技术成果转化，主导模型在快手短视频社区的集成落地（例如RAG增强系统、多模态Agent任务规划），并通过API封装与服务化支撑业务创新；
- 跟踪前沿多模态大模型后训练技术的前沿演进，探索SFT自动化迭代、轻量化RLHF以及reward model一体化等创新技术方向，形成专利或顶会论文提升团队技术影响力。 1、​​硕士及以上学历，计算机科学、人工智能、自动化、数学等相关专业优先； 2、精通多模态任务设计范式（如视觉思维链、跨模态推理链），具备CoT提示工程、Reward Model设计经验，掌握合成数据生成、多模态数据清洗、数据质量评估工具开发。熟练应用PyTorch+DeepSpeed、Megatron-LM等分布式框架，精通QLoRA、DPO、RLHF以及PPO/GRPO/DAPO等后训练技术； 3、主导过​​千亿级模型的后训练全流程​​，包括SFT数据构建、RM训练、RLHF/DPO对齐优化、端侧量化部署。具备​​丰富的多模态数据治理经验​​，如构建百万级图文对齐数据集，或设计视频-文本跨模态评测基准； 4、热衷于​​推动多模态大模型、生成式AI、Agent等前沿技术在快手生态中的发展与应用，能系统性拆解业务需求为​并制定可量化的任务目标。具备跨团队协同能力，能够有效推动算法、数据、工程团队建立标准化数据交付流程。

## At a glance
- Posted by the employer: Not stated
- First seen by OpenNash: Oct 19, 2026
- Last checked: Oct 19, 2026
- Listed for: Under a day
- Location: Beijing[Apply on Kuaishou Technology’s careers site →](https://zhaopin.kuaishou.cn/#/official/social/job-info/26279)

## More jobs at Kuaishou Technology

### [(大模型专项）多模态大模型算法工程师](https://jobs.opennash.com/companies/kuaishou-technology/jobs/98edf462c05648daadcf/)

Beijing

### [AI Agent 开发实习生（电商在线推荐引擎）](https://jobs.opennash.com/companies/kuaishou-technology/jobs/5945a85b9a833fcbb5f8/)

Beijing

### [AI Agent 策略实习生](https://jobs.opennash.com/companies/kuaishou-technology/jobs/b814e14328b750da25f6/)

Beijing

### [AI Agent算法专家](https://jobs.opennash.com/companies/kuaishou-technology/jobs/0e36be4e740de43e02ad/)

Beijing

### [AI Agent算法专家（智能创作方向）](https://jobs.opennash.com/companies/kuaishou-technology/jobs/41a78c4c892cbbd1bc50/)

Beijing

OpenNash lists jobs from employers’ own careers sites and does not handle applications.
Dates come from the employer where stated; “first seen” and “last checked” are our own observations.

OpenNash Jobs · Official listings with dates and links to the employer. [Weekly hiring report](/reports/hiring/) · [Methodology](/methodology/)