招聘城市：北京，杭州
…招募对大模型、智能体（Agent）与多模态内容理解具有浓厚研究兴趣的实习生，围绕 Agentic 内容理解体系中的核心算法问题开展前沿研究。岗位聚焦于真实复杂场景中的内容理解问题，探索基础模型、智能体架构与自进化机制在多模态任务中的能力边界与方法创新。
该岗位强调研究深度、问题导向与学术产出，适合希望在实习期间持续推进研究课题、沉淀论文成果，并将研究工作与实际高价值场景相结合的同学。
【研究方向和工作职责】
1. 研究 Agentic RL 长程任务强化学习方法，提升稳定性、一致性与任务完成能力。
2. 多模态检索与 Embedding 建模，提升多模态语义匹配和召回效果。
3. 探索 Agent 自进化机制，提升智能体自我评估、修正与持续…
招聘城市：贝尔维尤
…Business Unit
What the Role Entails
Research Internship – AgentTencent AI Lab is dedicated to advancing cutting-edge AI technologies, with a particular focus on innovative breakthroughs in large foundation models. The lab's long-term ambition is to drive the development of Artificial General Intelligence (AGI), and ultimately, Artificial Superintelligence (ASI). We are currently seeking research interns in Agents, with an emphasis on developing stable and efficient scalable RL framework and algorithms for our Seattle area office located at Bellevue WA for the year 2026.
Every research intern will work with researchers on a research project aimed at attacking one of the core problems on the design and optimization of scalable RL frameworks and algorithms for self-improving Agents with various environments, including Web Agent, GUI Agent. Research areas include but are not limited to RL Algorithms, Reward Modeling, World Models, GUI Agent, and Web Agent.
Who We Look For…
招聘城市：北京，上海，杭州
…旨在探索一种基于“教学—纠正”闭环的交互式进化审核 Agent 系统，致力于打破传统审核模型对静态规则与大规模标注样本的依赖，实现“规则—策略”的自动生成优化反馈闭环。
不同于通用 Agent，该系统强调在复杂、多变的国际化内容审核场景中，通过引入 Policy Maker 的实时干预与 Rule Set 的动态反馈，实现从“规则驱动”向“智能进化”的范式迁移。研究核心在于构建一套集成经验进化学习、在线学习及 RLRF（基于反馈的强化学习）的复合技术架构。关键问题包括：如何将抽象的审核政策（Policy）自动化解析为可执行的 Agent 策略链路，如何在跨语言、跨文化背景下构建具备自主学习能力的 Agent 基座，以及如何在极度稀疏的违规样本中利用小样本（Few-shot…
招聘城市：北京，上海，杭州
岗位职责：
探索一种自进化Agent系统，使Agent能够在真实环境中通过持续交互不断优化自身能力。不同于静态模型训练，该方向强调“生成—执行—评估—更新”的闭环过程。
关键问题包括：如何基于真实任务构建可靠的反馈信号，如何从稀疏成功案例中提取可泛化策略，以及如何避免自我强化中的分布偏移与错误积累。
平台提供多场景Agent执行环境与完整轨迹数据，使得自进化机制可以在真实任务中验证。该方向旨在推动Agent从“被动能力载体”向“主动学习系统”转变，是实现长期智能演进的重要路径。
任职要求：
1、不限年级，本科及以上在读，计算机/人工智能/软件工程等相关专业优先；
2、扎实的编程能力和算法功底，熟练掌握Python/C++/Java等至少一种…
招聘城市：北京，上海，杭州
岗位职责：
关注如何将RL引入工业级Agent平台系统，直接对“规划—执行—反馈”的完整轨迹进行优化。
研究重点包括：如何构建trajectory-level reward、如何在工具调用与多步推理中进行credit assignment，以及如何在高成本环境下进行高效的offline/online混合训练。平台提供真实任务环境与多样化Agent执行数据，使研究从离线benchmark走向真实交互场景。该方向有望推动RL从模型对齐走向复杂任务能力学习，形成新一代Agent优化范式。
任职要求：
1、不限年级，本科及以上在读，计算机/人工智能/软件工程等相关专业优先；
2、扎实的编程能力和算法功底，熟练掌握Python/C++/Java等至少一种编程语言；
3、扎实的机器学习/深度学习理论基础，有大规模推荐系统、计算广告、搜索引擎…
招聘城市：新加坡
…etc.)
- Contribute to papers, technical reports, and open-source projects
- Collaborate with cross-functional teams on research prototypes
Who We Look For
Currently pursuing a PhD or Master’s in Computer Science, AI, Machine Learning, or related fields
Strong background in deep learning and machine learning fundamentals
Solid programming skills in Python and PyTorch/JAX
Experience with at least one of:
- Vision–language models
- Large language models
- Video understanding/generation
- Reinforcement learning or imitation learning
- Strong problem-solving and research skills
Publications at top conferences (CVPR, ICCV, NeurIPS, ICLR, ACL, etc.)
Experience training large models or working with distributed systems
Experience with multimodal datasets and evaluation benchmarks
Familiarity with:
- Transformer architectures and scaling laws
- Multimodal alignment (contrastive learning, instruction tuning)
- Agent training (RLHF/RLAIF, planning, tool use)
- Synthetic data generation or simulation environments
- Experience with long-context training or memory mechanisms
Equal Employment Opportunity at Tencent
As an equal opportunity…
招聘城市：北京，上海，杭州
岗位职责：
本课题的研究目标是针对多Agent协同场景构建基于课程学习与分层强化学习的RL框架，从优先级经验回放(PER)、分布式经验复用和Actor-Critic异步计算优化等角度，攻克多目标冲突下的样本利用率低效问题。
该技术旨在突破传统RL训练在复杂任务（如小红书社区点点RL训练任务）中收敛慢、资源消耗高的瓶颈，实现训练效率提升3倍以上，支撑Agent服务快速迭代上线需求。
任职要求：
1、不限年级，本科及以上在读，计算机/人工智能/软件工程等相关专业优先；
2、熟悉Linux/Unix平台上的C++编程，熟悉网络编程-多线程编程，有良好的编程习惯；
3、熟悉其中一种主流的深度学习训练或推理框架（TensorFlow / PyTorch / Onnx / TensorRT…
招聘城市：伦敦
…learning frameworks and multi-agent collaboration algorithms to improve coordination and optimization across intelligent systems.
Who We Look For
Qualifications
• Master’s or PhD student in Computer Science, Mathematics, Statistics, Artificial Intelligence, or a related field.
• Strong foundation in mathematics (linear algebra, probability, statistics) and excellent problem-solving skills.
• Solid understanding of machine learning and deep learning algorithms (e.g., neural networks, decision trees, SVMs, reinforcement learning), with hands-on implementation experience.
• Proficiency in Python and familiarity with mainstream ML/DL frameworks such as TensorFlow and PyTorch, with the ability to independently develop and train models.
Preferred Qualities:
• Deep curiosity and passion for autonomous systems, reasoning, and multi-agent collaboration.
• Ability to bridge theoretical research and practical application.
• Experience with reinforcement learning, tool-augmented LLMs, or agentic workflows is a strong plus.
• Proficiency in both English and Mandarin will be a plus.
• Minimum 6-month full-time internship preferred.
#LI…
招聘城市：北京，上海，杭州，深圳
岗位职责：
本课题聚焦全模态Agent（GUI操作、代码生成、网页导航）在长程交互任务领域的算法研究。旨在解决Agent执行跨越数百至数千步的复杂任务时，传统强化学习仅依赖终态奖励信号，导致信用分配路径过长、梯度信号衰减，策略优化难以收敛的问题。研究方向包括：设计层次化时间抽象信用分配机制，缓解长程任务中flat policy的优化不稳定问题；设计验证跨模态可验证奖励机制，抑制Reward hacking对训练过程的干扰，实现全模态长程Agentic RL 稳定收敛。该研究成果将在WebArena、SWE-bench等主流评测基准上验证方法的有效性，应用于公司内社区生态Agent基座、AI跨模态深度搜索等业务场景，并集成至自研开源的强化学习引擎Relax，增强业界…
招聘城市：贝尔维尤
…extensively at top conferences and journals.Research Internship – AgentTencent AI Lab is dedicated to advancing cutting-edge AI technologies, with a particular focus on innovative breakthroughs in large foundation models. The lab's long-term ambition is to drive the development of Artificial General Intelligence (AGI), and ultimately, Artificial Superintelligence (ASI). We are currently seeking research interns in Agents, with an emphasis on developing stable and efficient scalable RL framework and algorithms for our Seattle area office located at Bellevue WA for the year 2026.
Every research intern will work with researchers on a research project aimed at attacking one of the core problems on the design and optimization of scalable RL frameworks and algorithms for self-improving Agents with various environments, including Web Agent, GUI Agent. Research areas include but are not limited to RL Algorithms, Reward Modeling, World Models, GUI Agent, and Web Agent.
Who We Look For…
招聘城市：北京，上海
岗位职责：
1、横评端侧开源 LLM（Gemma/Qwen/Phi/Llama），建立准确率 × 延迟 × 内存的 Pareto 基线
2、研究 Tool-Use 场景下的量化精度损失与补回方案（GPTQ/AWQ + LoRA 微调）
3、设计端云 Hybrid 路由策略：哪些意图端侧跑、哪些上云
4、构建端侧 Agent 评测 Benchmark
任职要求：
1、硕士/博士在读，熟悉 Transformer 架构与 LLM 推理流程
2、熟悉 PyTorch，了解 LiteRT / llama.cpp / ExecuTorch 等推理框架至少一种
3、有模型量化或蒸馏实践经验
4、加分：Android NDK 经验 · NPU 编程 · MLSys/MobiSys 发表
招聘城市：北京，上海
岗位职责：
1、横评端侧开源 LLM（Gemma/Qwen/Phi/Llama），建立准确率 × 延迟 × 内存的 Pareto 基线
2、研究 Tool-Use 场景下的量化精度损失与补回方案（GPTQ/AWQ + LoRA 微调）
3、设计端云 Hybrid 路由策略：哪些意图端侧跑、哪些上云
4、构建端侧 Agent 评测 Benchmark
任职要求：
1、硕士/博士在读，熟悉 Transformer 架构与 LLM 推理流程
2、熟悉 PyTorch，了解 LiteRT / llama.cpp / ExecuTorch 等推理框架至少一种
3、有模型量化或蒸馏实践经验
4、加分：Android NDK 经验 · NPU 编程 · MLSys/MobiSys 发表
招聘城市：洛杉矶
…wide range of services and resources to our network of developers and partner studios around the world to help them unlock the true potential of their games.
What the Role Entails
As a game research assistant, you will work closely with your mentor & other (senior) researchers to gather, analyze and interpret data that will directly impact our game development and marketing strategies. This is an exciting opportunity for anyone looking to gain hands-on experience in consumer insights as well as the gaming industry.
Job Description
Assist in all phases of quantitative and qualitative research process (surveys, focus groups, interviews), including study design, questionnaire drafting, fielding, data analysis & reporting.
Assist in quality control processes at each stage of the research, including proof-reading, checking survey links & data cleaning.
Assist with data visualization and presenting findings in a clear & concise manner.
Conduct desk research to support research projects including competitor analysis…
招聘城市：北京，上海，杭州
岗位职责：
传统审核大模型通常采用SFT的方式逼近人审对审核规则的识别精度，此时人工执行质量和规则合理性则成为机审体系性能上限。
本课题通过RLVR和Multi-Agent的方式，构造机审判别Agent与规则生成Agent的博弈学习，以对抗上升的方式不断提升审核规则的完备性以及相应机审识别的准召，使得机审可以突破人工上限，实现大模型智能在审核象限的涌现和“Aha moment”。
任职要求：
1、不限年级，本科及以上在读，计算机/人工智能/软件工程等相关专业优先；
2、优秀的代码能力、数据结构和基础算法功底，熟悉Python等至少一门编程语言；
3、熟悉大模型领域尤其是强化学习相关研究工作和算法，有大模型…
招聘城市：北京，杭州，上海
岗位职责：
传统的AI搜索依然基于RAG框架，少有的几个Agent框架也只涉及QueryPlanning，距离真实解决搜索中的实际问题还相距很远，例如做旅游攻略、做行业研究报告等等。我们判断，虽然当下LLM已经大范围的用于搜索领域，但是下一代的搜索技术变革一定是基于Agent的。本课题旨在研究基于Agent框架的基座模型。
任职要求：
1、不限年级，本科及以上在读，计算机/人工智能/软件工程等相关专业优先；
2、扎实的编程能力和算法功底，熟练掌握Python/C++/Java等至少一种编程语言；
3、扎实的机器学习/深度学习理论基础，有大规模推荐系统、计算广告、搜索引擎等核心算法项目经验；
4、在顶级学术会议或期刊发表论文，或ACM编程竞赛/机器学习等…
招聘城市：北京，上海
岗位职责：
1、参与 Agentic 系统的研究与原型开发，推动前沿技术在小红书场景的落地探索
2、设计并实现 Agent 框架的核心模块（规划、推理、记忆、工具调用）
3、构建实验环境，包括 Agent 模拟器、评测基准和数据 pipeline
4、设计 Agent UI DSL（领域特定语言），让 LLM 输出可渲染的结构化 UI 描述
任职要求：
1、计算机科学、人工智能相关专业硕士或博士在读
2、有信息检索（IR）、推荐系统、NLP 或多智能体系统相关的研究经历，有顶会论文发表经历者优先
3、扎实的编程能力（Python 必须，熟悉 PyTorch/HuggingFace/LangChain 等至少一个框架）
4、熟悉 LLM 的 prompting、fine-tuning 或 Agent 框架（如 AutoGPT、CrewAI、LangGraph 等）
5、具备独立阅读和复现前沿论文的能力
6、良好的英文…
招聘城市：新加坡
岗位职责：
Business Unit
What the Role Entails
1.Conduct research and development on multimodal processing algorithms and models, including but not limited to image and video understanding and generation, as well as alignment and integration of multimodal information with textual data;
2. Design and optimize existing algorithms to improve performance and accuracy, ensuring a high-quality user experience;
3. Conduct in-depth research on and stay up to date with cutting-edge technologies in multimodal learning, NLP, CV, and related fields, and promptly apply new technologies to products.
4. Explore and develop reinforcement learning algorithms and frameworks, including but not limited to policy optimization, reward modeling
Who We Look For
1. Bachelor’s degree or above in Computer Science, Information Engineering, Pattern Recognition, Artificial Intelligence, or related fields;
2. Proficient in fundamental algorithms and applications related to computer vision and image processing; familiar with at least one deep learning…
招聘城市：北京，上海
…1. 参与广告外投场景的 AI Agent 系统建设，完成基于 Agent 的投放策略、自动化巡检、预算分配等模块的研发工作。
2. 参与素材评估与召回算法优化，基于大模型与机器学习技术，提升素材质量与匹配效率。
3. 跟进 Agent 前沿技术（Memory、Self-improvement、Tool Use 等），完成技术调研与原型验证。
任职要求：
1. 硕士及以上学历在读，计算机/AI相关专业，每周出勤 4 天以上，实习期 3 个月以上。
2. 动手能力强，有扎实的编程基础，熟练使用 Vibe Coding 工具，能快速将想法转化为可运行代码。
3. 对 AI Agent 有热情，日常使用或折腾过各类 LLM/Agent 工具，对"用 AI 解决实际问题"有持续好奇心。
4. 有大模型/Agent 相关项目经验、竞赛获奖或开源贡献…
招聘城市：伦敦
…connection, and accessibility. Level Infinite also provides a wide range of services and resources to our network of developers and partner studios around the world to help them unlock the true potential of their games.
What the Role Entails
1. Assist in the development and improvement of AI agents, focusing on marketing and user acquisition.
2. Collaborate with data scientists and engineers to enhance the performance and efficiency of agent tasks.
3. Support cross-functional teams (marketing, operations, product) to optimize agent-driven strategies.
Who We Look For
1. Currently enrolled in a CS/EE degree program (undergraduate or postgraduate). Must be available for at least 6 months of internship.
2. Strong proficiency in programming languages such as Python and SQL.
3. Experience or interest in working with AI-driven systems is a plus.
4. Familiarity with basic marketing and user acquisition concepts is a plus.
#LI-RL1
Equal Employment…
招聘城市：北京，上海，杭州
岗位职责：
本课题聚焦于提升大模型在上百轮交互的超长程 Agent 任务中的自主执行能力，目标场景涵盖 CLI/GUI Computer Use、Software Engineering、科研与专业任务等，并致力于让模型从被动响应指令转变为能够主动推进任务的 Proactive Agent。当前面临的核心挑战贯穿训练与推理两侧：在训练侧，缺乏能够覆盖真实世界复杂性与多样性的 RL 训练环境，现有环境难以模拟长程任务中工具调用、状态变迁与多步依赖的真实分布；稀疏的结果奖励无法为上百步的中间过程提供有效训练信号，如何设计面向长程任务的 Reward Signal 与 Credit Assignment 机制是关键瓶颈；在推理侧，模型在长时执行中面临目标漂移、错误累积与上下文认知负载持续…
