关于我
About
现任镁佳科技智能座舱产品经理,参与长安深蓝座舱 AI 方向,围绕 MegaClaw、地图 Agent 与语音交互梳理用户场景、任务流程和产品边界。
此前在猎豹移动集团 EasyClaw负责 AI Agent 工作流,独立全栈开发多 Agent 运营系统,覆盖 6 条业务线、节省 6 名运营人力,并推动增长组运作产出提升 400%。
拥有英国约克大学人机交互硕士背景,并通过 CDS 车辆设计训练连接用户研究、整车 Package 与智能座舱 HMI。
Currently a Digital Cockpit Product Manager at Megatronix, contributing to Changan DEEPAL cockpit AI across MegaClaw, Navigation Agent and voice interaction.
Previously at Cheetah Mobile Group's EasyClaw, designed AI Agent workflows and independently delivered a full-stack multi-agent operations system across six business lines, saving six operations FTEs and increasing growth-team output by 400%.
Holds an MSc in Human-Centred Interactive Technologies from the University of York, with CDS vehicle-design training connecting user research, vehicle package constraints and digital-cockpit HMI.
工作 & 项目
Experience
- 座舱 AI 产品规划:围绕车内高频出行场景拆解用户需求、交互路径与功能边界
- MegaClaw 产品协同:梳理座舱 AI Agent 能力边界、任务流与交互触点,参与概念方案、原型设计、需求评审与研发沟通
- 地图 Agent 场景设计:覆盖导航、POI、路线规划与出行决策,设计“意图识别—信息推荐—操作闭环”的车载交互链路
- 语音体验优化:分析多轮对话、任务中断与误识别问题,输出优化策略与验收要点
- Cockpit AI Product Planning: Define user needs, interaction flows and functional boundaries for high-frequency in-car mobility scenarios
- MegaClaw Product Collaboration: Map Agent capabilities, task flows and interaction touchpoints; support concept development, prototyping, requirements reviews and engineering alignment
- Navigation Agent Experience: Design a closed-loop journey across navigation, POI, route planning and travel decisions, from intent recognition to recommendation and action
- Voice Experience Optimisation: Analyse multi-turn dialogue, task interruption and misrecognition; define optimisation strategies and acceptance criteria
- AI Agent 工作流设计:负责自动化工作流设计,方案获 CEO 傅盛选中并推动内部 AI Native 实践,增长组运作产出提升 400%
- 多 Agent 阵列运营系统:单人全栈开发,覆盖 6 条业务线,为增长组节省 6 名运营人力
- 用户增长与商业化:设计龙虾社区与 EasyClaw 分销平台,覆盖 UGC、裂变分销与用户转化
- AI Agent Workflow Design: Designed automation workflows selected for wider adoption by CEO Fu Sheng, helping catalyse an internal AI-Native shift and increasing growth-team output by 400%
- Multi-Agent Operations System: Independently delivered the full-stack system across six business lines, saving the equivalent of six operations FTEs
- User Growth & Commercialisation: Designed the Lobster Community and EasyClaw affiliate platform across UGC, viral distribution and user conversion
- UI 设计与视觉优化:负责 20 个界面模块的 UI 资源交付
- 交互与动效协作:与开发、策划团队协作,参与 12 个核心系统的 UI 动效与交互设计
- 界面结构优化:提出信息层级与布局优化方案,使核心页面交互清晰度评分提升 15%
- UI Design & Visual Optimisation: Delivered UI assets for 20 interface modules
- Interaction & Motion Collaboration: Partnered with engineering and game-design teams across 12 core systems
- Information Architecture: Proposed hierarchy and layout improvements, increasing interaction-clarity scores by 15%
- 数据清洗与分析:使用 SPSS、Python 处理与分析 54 万条使用数据,挖掘用户行为与潜在需求
- 数据驱动决策:与产品和开发协作,推动 3 项核心功能交互优化,用户任务完成率提升 18%
- 用户体验提升:基于数据优化界面布局与交互逻辑,整体用户满意度评分提升约 10%
- Data Cleaning & Analysis: Used SPSS and Python to analyse 540,000 usage records and identify user behaviours and unmet needs
- Data-Informed Product Decisions: Partnered with product and engineering to optimise three core features, increasing task completion by 18%
- UX Optimisation: Supported data-led improvements to layout and interaction logic, lifting user-satisfaction scores by approximately 10%
- 构建用户问卷体系,调研用户明确核心功能:地图导航、兴趣点推荐、活动提示
- 引入游戏化机制:签到奖励、探索成就系统,设计以「任务驱动」和「探索成就」为核心的界面
- 引入行为动机设计原则,提升用户在城市探索中的沉浸感
- 组织并参与用户测试,收集反馈进行迭代优化
- Designed a user survey framework to identify core features: map navigation, POI recommendations, and event alerts
- Introduced gamification mechanics — check-in rewards and exploration achievements — grounded in BJ Fogg's behaviour model
- Applied behavioural motivation design principles to deepen user immersion in city exploration
- Organised and facilitated two rounds of usability testing; iterated design based on feedback
- 用户与场景分析:分析城市出行需求、驾驶者视野、乘坐姿态与操作触点,提炼座舱功能布局
- 车辆与座舱方案:围绕外观比例、内饰分区、HMI 触点与空间可行性建立跨域连接
- 高强度迭代:每日 14 小时训练草图、线条、比例与功能完整性表达
- User & Scenario Analysis: Examined urban-mobility needs, driver visibility, seating posture and control touchpoints to inform cockpit layout
- Vehicle & Cockpit Concept: Connected exterior proportion, interior zoning, HMI touchpoints and spatial feasibility
- Rapid Iteration: Practised sketching, line quality, proportion and functional expression for 14 hours per day
教育背景
Education
YORK
of York
人机交互 · 硕士
研学纽北、古德伍德等欧洲汽车文化
专攻 AI 出行方向 · App 原型获推荐至英国议会展示
Explored European automotive culture at the Nürburgring and Goodwood
Focused on AI mobility · app prototype recommended for presentation to the UK Parliament
GZHU
University
广州大学华软软件学院
Huaruan Software Institute, GZHU
四个广东省级竞赛奖项
创立三个大型学生组织
Four Guangdong provincial competition awards
Founded three large student organisations