🇨🇳 中文版 🇬🇧 English Version

[招聘] 大模型算法工程师 / 研究科学家

阿里巴巴淘天集团 (Taobao and Tmall Group) 1688 | 多模态与产业智能团队

团队介绍

开放职位

1. 工业级 VLM 多模态底座

核心: 构建工业级 VLM 底座，解决复杂物理数据表征与对齐，提供高精度数字基建。

  • 工作: 工业级多模态底座全流程迭代；设计先进 VLM CPT, RLHF/RLAIF 多模底座；提出完善 数据合成 算法管线；重构海量品商 Embedding 与 Rerank 引擎，确保极高确定性的分发。
  • 要求: AI相关硕博，顶会核心贡献。深谙 VLM 架构与 Scaling Laws，模型直觉强。

2. 产业大模型与知识图谱

核心: 构建产业 LLM 底座，通过全链路训练与复杂推理，打造高精度产业知识图谱。

  • 工作: 主导大模型产业级 CPT、Post-train 与 RL；提升复杂逻辑推理能力；利用大模型抽取构建动态 知识图谱，并完善严苛的工业级 Evaluation 体系。
  • 要求: 扎实 NLP/LLM 基础，大规模 CPT 训练、信息抽取或推理优化经验丰富。

3. C2M Agentic 算法

核心: 优化电商垂域 Agent 基础设施，通过后训练与强化学习提升复杂环境高确定性推理。

  • 工作: 深度优化后训练算法（GRPO, PPO, SearchR1），提升模型 Tool-use 与 Deep Research 能力；构建高保真环境，攻坚 Multi-Agent RL 闭环。提升多种非标商品的 Agent2Agent（CLI）端到端 C2M/M2C 确定性。
  • 要求: 深谙 Alignment 算法。有设计美感，物理几何直觉好优先。

4. 2D/3D AIGC 与可制造性

核心: 突破数字生成到物理制造的算法瓶颈，实现从 AIGC 到工厂参数的无缝编译。

  • 工作: 研发前沿 2D/3D AIGC 算法；攻克 参数化重构与工业修复 难题，将视觉 Mesh 几何转换为高精度工业标准（CAD/CAM，满足打样公差）。
  • 要求: 2D/3D 视觉/GenAI 领域（Mesh, NeRF, Gaussian Splatting）深厚积累。有工业参数化或 CAD 内核研究背景优先。

🎓 研究型实习生 (Research Intern)

核心: 依托阿里真实商业场景与充足算力，在解决万亿级供应链的硬核难题中冲击顶会顶刊。

  • 工作: 深入四大核心方向。在百篇顶会经验的 Area Chair 亲自指导下，打通“业务落地-前沿研究-顶级发表”的闭环，将工业级算法突破转化为学术与开源影响力。
  • 要求: AI相关专业在读硕博。极强的代码工程能力与科研自驱力，兼具学术野心与 落地务实。有相关顶会发表经验优先，实习 3-6 个月以上。
工作地点： 杭州
开放层级： P5 - P8 均可 （从天才校招/顶尖实习生到资深算法专家全覆盖）
投递简历： liangding.liam@gmail.com / zuorui.dl@alibaba-inc.com
联系微信： Liam_DL1469
✦ ✦ ✦

[Hiring] Research Scientist / Algorithm Eng.

Alibaba Taobao and Tmall Group (1688) | Multimodal and Industrial Intelligence Team

About Us

Open Positions

1. Industrial VLM & Multimodal Foundation

Focus: Build industrial VLM foundations to resolve complex physical data representation and alignment.

  • Work: End-to-end VLM iteration; design advanced CPT & RLHF/RLAIF pipelines; build Data Synthesis workflows; rebuild Embedding & Rerank engines for massive B2B data.
  • Reqs: Ph.D./MS in AI with core contributions in top conferences. Deep intuition for VLM architectures and Scaling Laws.

2. Industrial LLM & Knowledge Graph

Focus: Construct LLM foundations and high-precision knowledge graphs via complex reasoning.

  • Work: Lead large-scale CPT, Post-train & RL; enhance logical reasoning; extract dynamic Knowledge Graphs; build rigorous industrial-grade Evaluation metrics.
  • Reqs: Solid NLP/LLM background. Expertise in large-scale CPT, information extraction, or reasoning optimization.

3. C2M Agentic Algorithms

Focus: Optimize e-commerce Agent infrastructure and highly deterministic reasoning via post-training and RL.

  • Work: Deeply optimize GRPO/PPO/SearchR1 for Tool-use and Deep Research; build high-fidelity environments and Multi-Agent RL loops. Enhance end-to-end C2M/M2C determinism via A2A (CLI) interactions.
  • Reqs: Expertise in Alignment algorithms. Strong geometric/physical intuition and design aesthetics are highly preferred.

4. 2D/3D AIGC & Computational Manufacturing

Focus: Bridge digital generation and physical manufacturing (seamless AIGC to factory parameters).

  • Work: Develop SOTA 2D/3D AIGC algorithms; tackle Parametric Reconstruction & Repair to convert visual meshes into CAD/CAM standards (meeting prototyping tolerances).
  • Reqs: Deep expertise in Vision/GenAI (Mesh, NeRF, 3DGS). Background in CAD kernel or industrial parametric algorithms is a major plus.

🎓 Research Intern

Focus: Tackle hard industrial challenges using Alibaba's massive real-world data and compute.

  • Work: Dive into our 4 core areas. Mentored directly by Area Chairs (100+ papers) to translate real-world B2B breakthroughs into top-tier papers (ACL/NeurIPS/ICLR) and open-source impact.
  • Reqs: Enrolled MS/Ph.D. in AI. Balances academic ambition with pragmatic execution. Prior top-tier publications highly preferred. Available for 3-6+ months.
Location: Hangzhou, China
Levels: P5 - P8 (From brilliant interns/fresh grads to seasoned technical experts)
Apply: liangding.liam@gmail.com / zuorui.dl@alibaba-inc.com
WeChat: Liam_DL1469
Visitor map