Agent产品经理

职位描述

1. Product Planning & Roadmap Design Lead AI Agent product planning and requirement decomposition based on real business and user scenarios. Define the Agent roadmap, balancing capability ceilings, system stability, and delivery cadence.
2. Scenario Modeling & Requirement Abstraction Abstract complex business problems into executable Agent architectures. Define Agent roles, responsibility boundaries, and measurable performance metrics such as success rate, cost, latency, and stability.
3. Workflow & Capability Design Design chained and parallel Agent workflows. Establish standards for prompt design, memory mechanisms, and external tool invocation strategies. Collaborate closely with model engineering, data, and front-end/back-end teams to translate designs into production-ready systems.
4. Evaluation & Continuous Optimization Design evaluation frameworks and human-in-the-loop strategies for Agents. Continuously optimize performance with a balanced focus on user experience, efficiency, and cost, and build sustainable iteration mechanisms.
5. Delivery & Commercialization Drive Agent deployment in real production environments and promote reuse and scaling across scenarios. Collaborate with operations, sales, and customer success teams to enable large-scale replication and commercial adoption. 1、产品规划与路线设计：面向真实业务/用户场景进行 AI Agent 产品规划与需求拆解， 制定Agent Roadmap，平衡能力上限、稳定性与交付节奏； 2、场景建模与需求抽象：将业务问题抽象为Agent 结构，定义 Agent 角色分工、技能边界与可衡量指标（成功率、成本、时延等）； 3、工作流与能力设计：设计链式/并行 Agent 工作流，构建 Prompt 规范、记忆、外部工具调用策略，协同模型工程、数据、前端/后端，将设计落地到可交付系统； 4、评估与迭代优化：设计评测人机协同策略，对 Agent 进行持续性能优化，兼顾体验、效率与成本，建立可持续迭代机制； 5、交付与商业落地：推动 Agent 在真实环境上线与复用扩散，与运营/销售/客户成功协同，完成规模化复制。

职位要求

1. Large Language Model & Agent Expertise Solid understanding of large language model fundamentals and mainstream enhancement strategies. Ability to make scenario-driven design and model decisions. Strong awareness of hallucination, drift, and uncertainty mechanisms, with practical mitigation approaches.
2. Model Ecosystem Knowledge Familiarity with mainstream model ecosystems (e.g., GPT, Qwen, Gemini, Claude, Llama), including differences in capability, cost structure, and security considerations.
3. Prompt & System Design Capability Deep understanding of structured prompt design principles, including constraint design, alignment, hierarchical task decomposition, and tool routing.
4. Product & Execution Experience 3+ years of experience in RPA, automation, or AI tooling products, or equivalent hands-on Agent development experience. Proven ability to work closely with engineering and data teams to drive complex capabilities into production.
5. Hands-on Agent Delivery Experience Experience delivering real-world Agent systems to production, with clearly defined evaluation metrics. Practical experience in prompt engineering, knowledge base construction, tool integration, and local or on-premise deployment. 1、具备大模型认知能力：理解大模型基本工作机制、理解主流增强策略、能基于场景做定制化判断、理解幻觉/漂移/不确定性产生机制，并能提出可执行的缓解措施； 2、熟悉主流模型生态（GPT / Qwen / Gemini / Claude / Llama 等）成本/安全上的差异； 3、理解 Prompt 结构化设计底层逻辑（约束/对齐/层次拆解/工具路由）； 4、3 年以上 RPA/ 自动化 / AI 工具类产品经验（或 Agent 实战经验），能与工程/数据团队高密度协作推进能力； 5、有真实 Agent 上线案例，并能给出评估指标，有 Prompt 工程 / 知识库 / 工具调用 / 本地部署实操经验。