{ "datasetId": "dragonai-ai-product-manager-course-syllabus-v1", "schemaVersion": "1.0.0", "createdAt": "2026-07-22", "updatedAt": "2026-07-22", "language": "zh-CN", "recordType": "first_party_course_syllabus", "sourceUrl": "https://course.dragonai.tech/elite#syllabus", "providerBrandZh": "烛龙智元 DragonAI", "providerLegalNameZh": "杭州烛龙智元科技有限公司", "courseNameZh": "AI 产品经理线下实战培训班 · 卓越班", "aggregates": { "moduleCount": 4, "syllabusEntryCount": 28, "coreLessonCount": 20, "guidedReviewCount": 6, "preparationCount": 1, "onlineSupportEntryCount": 1, "lessonsWithHandsOn": 20, "lessonsWithKnowledgePoints": 20 }, "modules": [ { "moduleIndex": 1, "moduleId": 1, "titleZh": "模块一：AI产品经理的基石", "subtitleZh": "基础认知与职业规划", "declaredLessonCount": 4, "actualLessonCount": 4, "lessons": [ { "lessonIndex": 1, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "大模型基础概念与技术原理", "categoryZh": "基础知识预习", "stageZh": "第1天（课前预习）", "durationZh": "课前录播", "coreTags": [], "descriptionZh": "深入剖析AI 2.0时代的核心技术，从Transformer架构到大模型训练三部曲（预训练、SFT、RLHF），揭示生成式AI的底层工作原理与无限潜能。", "knowledgePointsZh": [ "生成式AI与判别式AI范式革命", "Transformer自注意力机制", "Embedding与语义空间表征", "Token分词与处理原理", "大模型预训练与监督微调(SFT)", "人类反馈强化学习(RLHF)" ], "handsOnZh": "对比分析至少三款主流大模型产品（ChatGPT、Claude 3、文心一言等），从核心能力、局限性和产品定位等维度撰写分析报告" }, { "lessonIndex": 2, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "国产大模型巡礼与选型指南", "categoryZh": "基础知识预习", "stageZh": "第1天（课前预习）", "durationZh": "课前录播", "coreTags": [], "descriptionZh": "深入剖析国产顶尖大模型技术内核与能力边界，掌握从业务需求到成本效益的全方位模型选型策略，为打造卓越AI产品奠定坚实基础。", "knowledgePointsZh": [ "MoE架构与FP8低精度训练", "AI Agent统一能力框架", "模型推理性能指标(Latency/Throughput)", "多模态与超长上下文窗口技术", "国产主流模型技术栈(DeepSeek/Kimi/GLM/Qwen)", "AI产品经理模型选型策略" ], "handsOnZh": "设计一款垂直行业AI产品，撰写完整的模型选型报告，论证模型选择的合理性、潜在风险及应对策略" }, { "lessonIndex": 3, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "AI产品经理核心能力模型与就业规划", "categoryZh": "AI产品经理岗位及就业规划", "stageZh": "第2天（线下集中）", "durationZh": "线下课 2h", "coreTags": [], "descriptionZh": "深度剖析AI产品经理所需的技术架构认知、数据驱动思维和六维核心能力，掌握大厂青睐的实战方法论，明晰职业路径。", "knowledgePointsZh": [ "AI产品架构三层模型", "数据闭环与数据飞轮", "AI产品经理六维能力模型", "A/B测试与科学实验方法", "算法偏见与AI伦理治理", "模型优化目标定义" ], "handsOnZh": "分析3个AI产品经理JD，对照六维能力模型，撰写个人能力评估与提升计划" }, { "lessonIndex": 4, "entryType": "guided_review", "countsAsCoreLesson": false, "titleZh": "作业点评与学员答疑", "categoryZh": "AI产品经理岗位及就业规划", "stageZh": "第2天（线下集中）", "durationZh": "线下答疑", "coreTags": [], "descriptionZh": "导师逐一点评学员作业，针对模型选型报告和能力评估提供反馈与改进建议。", "knowledgePointsZh": [], "handsOnZh": null } ] }, { "moduleIndex": 2, "moduleId": 2, "titleZh": "模块二：AI核心技术与前沿框架", "subtitleZh": "技术深度 + 动手实战", "declaredLessonCount": 9, "actualLessonCount": 9, "lessons": [ { "lessonIndex": 1, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "RAG全景解析（一）：从基础到生产级", "categoryZh": "AI核心技术讲解（共4天）", "stageZh": "第3天（线下工作坊）", "durationZh": "线下课 3h + 实践 + 答疑", "coreTags": [ "RAG" ], "descriptionZh": "深入剖析RAG从基础到生产级的全景架构，掌握高质量知识库构建、高级检索优化与微调选型决策，赋能构建企业级智能问答与知识增强应用。", "knowledgePointsZh": [ "RAG架构演进：基础→高级→模块化", "生产级RAG五步流程：加载、分块、向量化、检索、生成", "高质量知识库构建：语义分块与向量数据库选型", "RAG与微调(Fine-tuning)技术选型框架", "高级RAG优化：查询转换与重排(Re-ranking)" ], "handsOnZh": "设计迷你RAG应用的知识库构建方案，涵盖信息源选择、文档加载、文本分块策略、元数据设计及向量数据库选型" }, { "lessonIndex": 2, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "RAG全景解析（二）：进阶优化与效果评估", "categoryZh": "AI核心技术讲解（共4天）", "stageZh": "第3天（线下工作坊）", "durationZh": "线下课 3h + 实践 + 答疑", "coreTags": [ "RAG" ], "descriptionZh": "探索高级RAG架构，掌握查询改写、智能路由与混合检索等核心优化技术，并运用RAGAS框架科学评估系统性能，构建生产级检索增强应用。", "knowledgePointsZh": [ "进阶RAG Pipeline架构设计", "查询改写与智能路由策略", "混合检索与重排序(Re-ranking)", "RAGAS效果评估体系", "假设性文档嵌入(HyDE)", "交叉编码器(Cross-Encoder)精排" ], "handsOnZh": "设计完整的RAG优化实验方案，包括场景选择、问题定义、技术选型（查询改写、路由、Reranker）和效果评估方法" }, { "lessonIndex": 3, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "Agent原理与多智能体协作", "categoryZh": "AI核心技术讲解（共4天）", "stageZh": "第3天（线下工作坊）", "durationZh": "线下课 3h + 实践 + 答疑", "coreTags": [ "Agent" ], "descriptionZh": "深入剖析AI Agent的核心工作原理与四大关键组件，掌握ReAct等主流设计模式，探索CrewAI、AutoGen等前沿多智能体协作框架。", "knowledgePointsZh": [ "Agent核心组件：规划、记忆、工具、行动", "ReAct推理与行动框架", "Plan-and-Execute设计模式", "Function Calling工具调用机制", "多智能体协作框架(CrewAI/AutoGen)" ], "handsOnZh": "分组设计一个多Agent协作的产品方案，定义Agent角色分工、通信协议和任务编排流程" }, { "lessonIndex": 4, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "多模态大模型技术与产品设计", "categoryZh": "AI核心技术讲解（共4天）", "stageZh": "第3天（线下工作坊）", "durationZh": "线下课 3h + 实践 + 答疑", "coreTags": [], "descriptionZh": "探索多模态AI核心技术，掌握视觉编码、跨模态对齐、扩散模型等关键技术，通过AI测肤、AIGC营销等实战案例学习前沿产品设计方法。", "knowledgePointsZh": [ "视觉编码器(ViT)与跨模态对齐", "多模态信息融合机制", "扩散模型(Diffusion Models)原理", "模型蒸馏与量化加速", "国产多模态模型(Kimi/通义千问/GLM)对比" ], "handsOnZh": "设计一个多模态AI产品概念（如AI菜谱生成助手），完成不少于500字的产品设计文档(PRD)初稿" }, { "lessonIndex": 5, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "AI在B/C端行业的落地场景与价值分析", "categoryZh": "AI落地场景与动手实战", "stageZh": "第4天（线下集中）", "durationZh": "线下课 2h + 实践 + 答疑", "coreTags": [], "descriptionZh": "剖析AI在金融、法律、教育、电商等行业的B/C端落地场景，掌握系统的场景发现框架与ROI价值评估方法。", "knowledgePointsZh": [ "B端AI应用价值框架", "C端AI应用价值框架", "大模型场景发现三步法", "AI项目ROI量化评估模型", "金融风控中的非结构化数据分析", "基于大模型的智能文档审查" ], "handsOnZh": "撰写一份\"AI+教育\"领域的迷你商业需求文档(Mini-BRD)，包含项目背景、目标用户、核心功能设计和商业价值评估" }, { "lessonIndex": 6, "entryType": "guided_review", "countsAsCoreLesson": false, "titleZh": "RAG+Agent作业点评与学员答疑", "categoryZh": "AI落地场景与动手实战", "stageZh": "第4天（线下集中）", "durationZh": "线下答疑", "coreTags": [ "RAG", "Agent" ], "descriptionZh": "导师逐一点评RAG知识库构建方案和Agent协作设计，深度反馈技术方案的可行性与优化方向。", "knowledgePointsZh": [], "handsOnZh": null }, { "lessonIndex": 7, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "动手实战：OpenClaw部署与应用", "categoryZh": "AI Agent平台实战", "stageZh": "第5天（线下实操）", "durationZh": "线下课 2h + 实践 + 答疑", "coreTags": [ "OpenClaw" ], "descriptionZh": "深入剖析GitHub十万星标项目OpenClaw，掌握其核心架构与产品策略，通过实战部署构建可自托管、连接多模型的个人AI助手。", "knowledgePointsZh": [ "OpenClaw中心辐射式架构", "Agent沙箱安全机制", "Docker化部署与模型连接", "Webhook与IM集成", "提示注入(Prompt Injection)攻防" ], "handsOnZh": "通过Docker实战部署OpenClaw，连接大语言模型，并集成到Slack等IM工具中，构建个人专属AI助手" }, { "lessonIndex": 8, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "动手实战：低代码Agent平台Dify/Coze", "categoryZh": "AI Agent平台实战", "stageZh": "第5天（线下实操）", "durationZh": "线下课 2h + 实践 + 答疑", "coreTags": [ "Agent" ], "descriptionZh": "深入掌握Dify与Coze两大低代码平台，通过可视化工作流与丰富插件，实战构建智能问答、客服机器人等企业级Agent应用。", "knowledgePointsZh": [ "Dify工作流编排引擎", "Coze的Bot构建与多渠道发布", "RAG知识库可视化构建", "Agent插件与工具调用", "Human-in-the-Loop人在环路机制", "低代码平台技术选型(Dify/Coze/FastGPT)" ], "handsOnZh": "基于Dify或Coze平台，构建一个智能助教机器人，包括创建知识库、设定角色、实现基础问答与特定工作流" }, { "lessonIndex": 9, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "LangChain核心拆解与AI编程入门", "categoryZh": "AI Agent平台实战", "stageZh": "第5天（线下实操）", "durationZh": "线下课 2h + 实践 + 答疑", "coreTags": [], "descriptionZh": "深入掌握LangChain核心组件与AI编程新范式，驾驭Cursor等AI原生工具，通过代码级理解力构建复杂大模型应用。", "knowledgePointsZh": [ "LangChain核心组件体系", "LangChain表达式语言(LCEL)", "LangSmith可观测性平台", "AI原生IDE(Cursor/Windsurf)", "Vibe Coding自然语言编程范式", "代码库上下文感知技术" ], "handsOnZh": "使用Cursor通过自然语言指令辅助编写并运行一个完整的LangChain应用，亲身体验Vibe Coding编程新范式" } ] }, { "moduleIndex": 3, "moduleId": 3, "titleZh": "模块三：AI产品全流程实战", "subtitleZh": "企业级项目 + 方法论", "declaredLessonCount": 8, "actualLessonCount": 8, "lessons": [ { "lessonIndex": 1, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "项目一：企业级AIGC内容平台", "categoryZh": "AI项目实战训练及拆解（超大型0-1项目）", "stageZh": "第6天（线下项目冲刺营）", "durationZh": "线下课 4h + 实践 + 答疑", "coreTags": [], "descriptionZh": "深入真实金融场景，从0到1构建企业级AIGC内容平台，掌握RAG、多模态内容生成、智能审核与效果评估全流程核心能力。", "knowledgePointsZh": [ "企业级AIGC解决方案设计", "RAG知识库构建与管理", "多模态内容生成引擎", "AI审核与校对工作流", "模型效果评估体系设计", "金融行业应用场景分析" ], "handsOnZh": "为\"智能客服问答\"场景设计迷你版PRD，包含背景目标、用户故事、核心功能详设（RAG知识库管理、AI辅助回答）及效果评估指标" }, { "lessonIndex": 2, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "项目二：企业级智能客服机器人", "categoryZh": "AI项目实战训练及拆解（超大型0-1项目）", "stageZh": "第6天（线下项目冲刺营）", "durationZh": "线下课 4h + 实践 + 答疑", "coreTags": [], "descriptionZh": "从0到1构建企业级智能客服，掌握意图识别、RAG等核心技术，学习AI辅助、自动化评测等创新模式，打造能创造商业价值的AI产品。", "knowledgePointsZh": [ "意图识别系统设计", "RAG检索增强生成实战", "向量数据库与知识库治理", "AI辅助与人机协同", "数据驱动与A/B测试", "智能客服工作流自动化" ], "handsOnZh": "选择一款主流App，逆向分析其智能客服的需求场景与用户流程，从产品经理视角提出至少3个迭代优化建议" }, { "lessonIndex": 3, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "AI产品经理核心工作方法论", "categoryZh": "AI产品经理核心工作方法", "stageZh": "第7天（线下集中）", "durationZh": "线下课 2h + 实践 + 答疑", "coreTags": [], "descriptionZh": "深入掌握AI产品从模型选型、效果评估到需求定义的完整方法论，学习构建自动化评测体系与数据驱动决策。", "knowledgePointsZh": [ "模型选型方法论", "模型效果评估体系", "AI产品PRD撰写规范", "A/B实验设计与分析", "自动化评测流水线", "数据驱动决策框架" ], "handsOnZh": "为L12/L13的实战项目撰写迷你版AI产品PRD，包含产品概述、AI任务定义、模型I/O规格、核心功能、效果评估指标和数据需求" }, { "lessonIndex": 4, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "AI产品商业化与GTM策略", "categoryZh": "AI产品经理核心工作方法", "stageZh": "第7天（线下集中）", "durationZh": "线下课 2h + 实践 + 答疑", "coreTags": [], "descriptionZh": "深入剖析AI产品的商业模式、成本结构与GTM策略，掌握从技术到市场价值的完整商业闭环，构建平台化生态思维。", "knowledgePointsZh": [ "AI产品商业模式(SaaS/API/订阅制)", "AI产品成本结构(算力/模型/数据)", "GTM市场进入策略", "AI产品定价策略(基于价值/分层)", "MaaS平台化战略", "AI产品生态建设与网络效应" ], "handsOnZh": "撰写一份AI产品的迷你商业计划书(Mini-BP)，涵盖产品概述、商业模式、成本预估及GTM策略" }, { "lessonIndex": 5, "entryType": "guided_review", "countsAsCoreLesson": false, "titleZh": "项目定义、架构图、流程图点评", "categoryZh": "AI产品经理核心工作方法", "stageZh": "第7天（线下集中）", "durationZh": "线下答疑", "coreTags": [], "descriptionZh": "导师逐一点评学员的项目定义文档、系统架构图和业务流程图，并提供产品逻辑与技术可行性的改进建议。", "knowledgePointsZh": [], "handsOnZh": null }, { "lessonIndex": 6, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "下一代AI协议：MCP与A2A深度解析", "categoryZh": "进阶拓展", "stageZh": "第8天（线下集中）", "durationZh": "线下课 2h + 实践 + 答疑", "coreTags": [ "MCP", "A2A" ], "descriptionZh": "深入解析下一代AI通信协议MCP与A2A，掌握构建模型与工具解耦、多智能体协作的开放式AI生态的关键，洞见可组合AI与Agent网络的产品新范式。", "knowledgePointsZh": [ "Model Context Protocol(MCP)核心架构", "Agent-to-Agent(A2A)协作协议", "模型与工具解耦设计", "多智能体系统(MAS)架构", "可组合AI(Composable AI)", "去中心化Agent商店" ], "handsOnZh": "设计一个基于MCP和A2A协议的\"智能旅行规划Agent\"产品概念，包括功能定义、工具/API设计、多Agent协作构想及架构图" }, { "lessonIndex": 7, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "AI伦理、安全与治理", "categoryZh": "进阶拓展", "stageZh": "第8天（线下集中）", "durationZh": "线下课 2h + 实践 + 答疑", "coreTags": [], "descriptionZh": "掌握数据隐私保护、算法公平性、模型可解释性与AIGC内容安全等前沿技术，学习头部大厂AI治理实践，构建负责任的AI产品设计体系。", "knowledgePointsZh": [ "数据隐私合规框架(GDPR/PIPL)", "算法偏见检测与缓解", "LIME/SHAP模型可解释性", "AIGC内容安全与过滤", "数字水印与深度伪造溯源", "负责任AI(RAI)设计原则" ], "handsOnZh": "撰写一份\"AI虚拟面试官\"产品的伦理风险评估简报，分析数据隐私、算法公平、可解释性等核心风险并提出应对策略" }, { "lessonIndex": 8, "entryType": "guided_review", "countsAsCoreLesson": false, "titleZh": "作业点评 / 学员答疑 / PRD初步点评", "categoryZh": "进阶拓展", "stageZh": "第8天（线下集中）", "durationZh": "线下答疑", "coreTags": [], "descriptionZh": "导师对MCP/A2A产品设计、伦理风险评估和PRD初稿进行逐一深度点评，为后续项目冲刺奠定基础。", "knowledgePointsZh": [], "handsOnZh": null } ] }, { "moduleIndex": 4, "moduleId": 4, "titleZh": "模块四：求职冲刺与毕业答辩", "subtitleZh": "项目打磨 + 面试突破", "declaredLessonCount": 7, "actualLessonCount": 7, "lessons": [ { "lessonIndex": 1, "entryType": "preparation", "countsAsCoreLesson": false, "titleZh": "完善个人项目与PRD", "categoryZh": "项目冲刺（共计2天）", "stageZh": "第9-10天（线下集中）", "durationZh": "课前录播 + 实践 + 答疑", "coreTags": [], "descriptionZh": "两天高强度项目冲刺期，计划在课程团队指导下推进 PRD、项目材料整理和路演材料打磨；实际完成情况取决于参与、作业与考评。", "knowledgePointsZh": [], "handsOnZh": null }, { "lessonIndex": 2, "entryType": "guided_review", "countsAsCoreLesson": false, "titleZh": "PRD正式点评与终评", "categoryZh": "项目冲刺（共计2天）", "stageZh": "第9-10天（线下集中）", "durationZh": "线下答疑", "coreTags": [], "descriptionZh": "导师对PRD终稿进行正式终评，从产品逻辑、技术方案、商业价值三个维度给出评分和改进建议。", "knowledgePointsZh": [], "handsOnZh": null }, { "lessonIndex": 3, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "Chat BI 项目案例实战", "categoryZh": "面试求职技巧 / 结课", "stageZh": "第11天（线下集中）", "durationZh": "线下课 2h + 实践 + 答疑", "coreTags": [], "descriptionZh": "掌握Chat BI核心概念与NL2SQL技术，学习企业级产品架构设计与竞品分析，通过从零构建项目案例获得实战经验。", "knowledgePointsZh": [ "NL2SQL核心技术原理", "Chat BI产品分层架构", "智能可视化与报表自动生成", "企业级Chat BI落地挑战与对策", "人机协作(Human-in-the-loop)机制", "\"Schema+规则+模型\"混合策略" ], "handsOnZh": "设计一个\"电商运营Chat BI\"的迷你产品原型(PRD)，包括产品概述、核心功能设计、数据源假设和非功能性需求" }, { "lessonIndex": 4, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "AI产品ROI评估方法", "categoryZh": "面试求职技巧 / 结课", "stageZh": "第11天（线下集中）", "durationZh": "线下课 2h + 实践 + 答疑", "coreTags": [], "descriptionZh": "掌握AI产品商业价值的核心评估体系，学习量化算力成本与业务收益的实战方法，培养数据驱动的决策能力。", "knowledgePointsZh": [ "成本-收益分析(CBA)框架", "算力成本量化模型", "核心收益量化方法", "A/B测试价值归因", "AI项目商业价值汇报" ], "handsOnZh": "选择一个具体行业，设计一款AI产品，撰写包含完整ROI分析的项目建议书" }, { "lessonIndex": 5, "entryType": "core_lesson", "countsAsCoreLesson": true, "titleZh": "大厂AI产品经理面试全解", "categoryZh": "面试求职技巧 / 结课", "stageZh": "第11天（线下集中）", "durationZh": "线下课 4h + 实践 + 答疑", "coreTags": [], "descriptionZh": "从简历优化、技术问题到案例实战，训练AI产品经理岗位常见的面试表达与复盘方法。", "knowledgePointsZh": [ "简历撰写与STAR法则", "RAG/Agent技术面试要点", "大模型核心评估指标", "AI产品商业模式设计题", "行为面试STAR-L模型", "大厂面试全流程拆解" ], "handsOnZh": "通过简历诊所、Agent设计、竞品分析等课堂实战，结合简历打磨、项目复盘PPT、行为面试题库等全面演练面试核心技能" }, { "lessonIndex": 6, "entryType": "guided_review", "countsAsCoreLesson": false, "titleZh": "毕业答辩与导师项目点评", "categoryZh": "面试求职技巧 / 结课", "stageZh": "第11天（线下集中）", "durationZh": "线下答疑", "coreTags": [], "descriptionZh": "学员进行毕业项目路演答辩，导师从产品思维、技术深度、商业价值和表达能力四个维度进行综合评定。", "knowledgePointsZh": [], "handsOnZh": null }, { "lessonIndex": 7, "entryType": "online_support", "countsAsCoreLesson": false, "titleZh": "线上课程支持（计划）", "categoryZh": "深度求职辅导及陪跑答疑", "stageZh": "结课后90天内（全线上陪跑）", "durationZh": "线上支持；形式与次数以正式协议为准", "coreTags": [], "descriptionZh": "结课后 90 天线上课程支持期内计划提供简历与项目材料反馈、模拟面试练习与复盘、公开岗位检索与申请策略及课程答疑；是否采用 1v1、可预约次数、具体形式与服务范围以报名前双方签署的正式协议为准，且不承诺面试、录取、就业、Offer、内推或特定结果。", "knowledgePointsZh": [], "handsOnZh": null } ] } ], "disclosureZh": "这是由课程提供方发布的第一方计划课程大纲，不是第三方评级；具体授课顺序、课时、讲师、练习与服务范围以报名前双方确认的最终课表和正式协议为准。", "resultBoundaryZh": "课程提供知识、项目、表达与求职能力训练，不承诺面试、录取、就业、Offer、内推或特定结果。", "measurementBoundaryZh": "本文件公开、抓取、下载或引用不等于搜索排名、品牌提及、外部链接、生成式 AI 引用或学习结果。", "license": "https://creativecommons.org/licenses/by/4.0/", "title": "DragonAI 卓越班 20 节核心课与完整课程安排 v1", "description": "公开 4 个模块、20 节核心课及 8 个预习、答疑或支持条目，逐项列出阶段、时长、知识点、练习与口径边界。", "url": "https://course.dragonai.tech/datasets/dragonai-ai-product-manager-course-syllabus-v1.json", "creator": "烛龙智元 DragonAI 教研团队" }