{"username":"guuumiho","header":{"name":"Tilda Pan","tagline":"FDE / AI Engineer / RAG Engineer focused on turning real business context into working AI solutions"},"personalInfo":{"email":"1011934252@qq.com","phone":"18258169339","location":"China","birthday":"Age 26"},"experience":[{"company":"待补充（财务 AI 项目）","role":"RAG 工程师","startDate":"2025.07","endDate":"2025.11","bullets":["参与财务智能体系统技术验证，围绕合同、报表、凭证等财务文档场景探索 RAG 检索问答、文档解析、结构化转换和语音交互能力。","参与模型选型与测试，对比 Qwen-VL、MinerU 等方案在文档解析精度、响应速度、多格式适配上的效果，并完成 Qwen3-VL-30B 私有化部署验证。","实现多格式文件向 JSON 的结构化转换，并同步开发重复文件检测功能，支持后续检索、审核和知识库处理链路。","定义实体与关系提取规则，完成多模态内容的实体识别、关系提取与结构化存储。","结合财务文档特性优化 Chunk 切分策略，设计关键词检索、语义检索、多条件组合检索等检索方式。","负责实时语音交互系统核心开发，实现流式语音转文字、声纹注册、说话人识别、回复打断和记忆系统管理等功能。","沉淀出对财务 RAG 适用边界的判断：严格固定流程更适合 RPA，而 AI/Agent 更适合自上而下的流程重构与复杂协作场景。"],"tags":["RAG","Qwen3-VL","MinerU","Document AI","JSON","Voice AI","Finance"]},{"company":"待补充（腾讯云 / FastGPT 项目）","role":"商务经理 / 云产品解决方案沟通","startDate":"2025.01","endDate":"2025.06","bullets":["主动拓展游戏、社交、医疗、传统行业等泛互行业云计算资源客户，了解行业动态、客户痛点和潜在商机。","负责 TAPD / FastGPT 用户接触、拜访、产品演讲、月度线下活动邀约和需求跟进。","通过上门拜访和线上沟通梳理企业客户业务场景、现有痛点与发展需求，协助确定需求优先级。","基于客户业务特性提供云产品选型建议和解决方案说明，用通俗语言讲解产品功能、技术优势和服务价值。","负责腾讯云、火山云、FastGPT 等产品的技术答疑，跟进客户试用和 POC 测试过程。","跟进客户使用情况，收集并反馈产品建议，处理试用和使用过程中的风险、纠纷和问题反馈。","月度完成指标多次超过老员工，成为组内 Top 1；积累客户拜访、需求理解、产品讲解、试用跟进和技术答疑经验。"],"tags":["ToB","Cloud","FastGPT","POC","Solution","Customer Discovery"]},{"company":"待补充（上汽大众项目）","role":"数据运维工程师","startDate":"2024.06","endDate":"2024.09","bullets":["负责上汽大众移动渠道端数据运维，校对多方车辆参数数据，维护车型版本、年款升级、价格、车型亮点、内饰、配置参数、选装包等数据展示。","使用 SQL 更新数据并校验各表关联，保障前端数据展示准确。","维护不同环境下的阿里云数据库、OSS、CDN 等资源。","与多个数据提供商协同工作，及时反馈错误数据并对齐字段颗粒度。","保障多源车辆参数、价格、配置、选装包等数据展示准确性，积累数据维护、环境资源和多方协同的问题定位经验。"],"tags":["SQL","Data Ops","Aliyun","OSS","CDN","Automotive"]},{"company":"待补充（上汽大众项目）","role":"测试工程师","startDate":"2024.06","endDate":"2024.09","bullets":["负责上汽大众推广传播程序的 Web 端、C 端与爬虫端功能测试、UI 测试、系统测试、Bug 提交和回归测试。","参与需求评审，主导编写并执行 1000+ 测试用例。","与产品经理、开发和用户沟通需求及预期效果，参与测试用例评审和问题确认。","建立从需求评审、测试用例、Bug 提交到回归测试的质量验证经验，形成较强的验证和复盘意识。","沉淀需求澄清、验证边界和 MVP 最小功能点判断能力，可支持 FDE 场景下的 demo 需求验证。"],"tags":["QA","Test Cases","Regression","Web","UI Testing","MVP"]},{"company":"待补充（宝格丽等零售品牌项目）","role":"品牌技术支持工程师","startDate":"2023.05","endDate":"2024.06","bullets":["以宝格丽项目为主提供二线技术支持，对接中国大陆及港澳台门店，并支持宝格丽、雅诗兰黛、LV、泸溪河等多个零售品牌。","支持 Beanstore、SAP、CRM、AD、OA 等多个核心业务系统，保障系统稳定运行。","日常监控 Cisco Meraki 等网络设备，配置出入栈规则，通过设备管理层实现企业网络流量管理与运维。","维护全国门店设备运行情况，为上门工程师提供二线支持，保障设备迭代、安装、拆卸顺利进行。","负责路由器、AP、交换机、打印机、销售仪器等软硬件设备管理与正常运转，并维护内部软硬件资产。","累计支持过万人次，涉及跨国团队，部分沟通语言为英语；多次获得集团月度奖金、集团 AI 应用大赛奖项，并带领团队赢得年度优秀团队奖项。","沉淀客户现场沟通、故障排查、跨团队协作和复杂问题拆解能力。"],"tags":["Technical Support","Retail IT","SAP","CRM","AD","Cisco Meraki","Cross-functional"]}],"education":[],"projectsRecent":[{"title":"本地 AI 问答与记忆实验工具","description":"面向高频 AI 使用者，探索高密度回复、关键信息本地沉淀、多模型切换、短期上下文和中期记忆提取机制。","url":"https://github.com/Guuumiho","tags":["LLM","Prompt","Memory","Multi-model"]},{"title":"Agent 项目源码辅助研读工具","description":"将源码研读拆解为目录树理解、核心入口定位、执行流程追踪和模块职责总结，帮助初学者理解 Agent 项目主流程。","url":"https://guuumiho.github.io/agent-code-learning","tags":["Agent","Source Reading","Frontend"]},{"title":"专注力训练小工具","description":"基于 Stroop 文字-颜色冲突法设计的专注训练工具，完成前端页面与部署。","url":"https://guuumiho.github.io/Stroop-Brain-Train","tags":["Frontend","Interaction","Cognitive Training"]},{"title":"任务引导工具","description":"面向多任务推进中的思维混乱问题，探索 OS 级任务感知、上下文捕捉、偏离提醒和及时方案提供。","url":"https://github.com/Guuumiho","tags":["AI Assistant","Tauri","Context","Product Prototype"]}],"projectsDetailed":[{"title":"本地 AI 问答与记忆实验工具","type":"Personal AI Tool","startDate":"2025","endDate":"Present","url":"https://github.com/Guuumiho","bullets":["通过系统 Prompt 设计强制简洁回复风格，迭代多个版本提示词，稳定输出高密度重点信息。","自动整理问题与答案，将重要信息以笔记形式本地沉淀，降低连续问答后的信息流失。","支持不同 AI 模型选择与调用，减少在多个平台之间手动切换的成本。","设计短期上下文 + 中期记忆提取机制，通过触发机制和摘要提取器模板提升连续问答体验。","针对临时问题设计隔离式单问模式，避免无关内容污染主线任务上下文。","针对网络和模型波动导致的调用失败，设计原模型重试、降级模型重试、备用 API 重试、失败提示和重新发送按钮。"],"tags":["LLM","Prompt Engineering","Memory","API","Fallback Strategy"],"externalLink":{"label":"GitHub","url":"https://github.com/Guuumiho"}},{"title":"Agent 项目源码辅助研读工具","type":"Learning Tool","startDate":"2025","endDate":"Present","url":"https://guuumiho.github.io/agent-code-learning","bullets":["面向 Agent 初学者“不知道先从哪里开始看源码、不知道某个文件作用、找不到主流程”等问题设计源码辅助研读工具。","将源码研读流程拆解为目录树理解、核心入口定位、执行流程追踪、模块职责总结。","帮助用户熟悉 Agent Loop、Context、Provider、Memory、Tool Calling 等 Agent 项目核心概念。","使用 nanobot 项目作为测试样例，完成前端部署体验入口。"],"tags":["Agent","Code Reading","Context","Tool Calling","Frontend"],"externalLink":{"label":"Live Demo","url":"https://guuumiho.github.io/agent-code-learning"}},{"title":"专注力训练小工具","type":"Frontend Tool","startDate":"2025","endDate":"Present","url":"https://guuumiho.github.io/Stroop-Brain-Train","bullets":["使用文字-颜色冲突的 Stroop 法设计专注训练工具。","完成前端页面与部署，通过文字与颜色冲突交互，引导用户进行注意力和反应训练。"],"tags":["Frontend","Stroop","Interaction Design"],"externalLink":{"label":"Live Demo","url":"https://guuumiho.github.io/Stroop-Brain-Train"}},{"title":"任务引导工具","type":"AI Assistant Prototype","startDate":"2025","endDate":"Present","url":"https://github.com/Guuumiho","bullets":["面向多任务同时推进导致的思维混乱问题，设计轻度任务引导工具。","探索 OS 级应用形态，让 AI 感知从对话框延伸到用户正在进行的任务上下文。","设计自动捕捉用户应用内容的能力，减少用户反复说明遇到困难时背景信息的成本。","针对提前过度优化、任务走向偏离、反复卡住的难题等场景设计轻度提醒和方案提供机制。"],"tags":["AI Assistant","Tauri","Context Awareness","Product Thinking"],"externalLink":{"label":"GitHub","url":"https://github.com/Guuumiho"}}],"skills":[{"name":"AI / LLM","items":["Model API 调用","Prompt 调优","多模型切换","上下文记忆","失败兜底策略","Agent Loop","Context","Provider","Memory","Tool Calling"]},{"name":"RAG / Knowledge Base","items":["多格式文档解析","结构化转换","Chunk 切分","关键词检索","语义检索","多条件组合检索","实体与关系抽取","多模态内容处理"]},{"name":"Engineering","items":["SQL 基础","数据维护","表关联校验","阿里云数据库","OSS","CDN","React","Tauri","Demo / Prototype Development"]},{"name":"Business / Field","items":["ToB 客户沟通","产品讲解","需求梳理","客户拜访","POC 跟进","技术答疑","客户现场支持","跨团队协作"]},{"name":"Quality / Operations","items":["系统性问题排查","测试用例设计","回归测试","Bug 提交","企业系统支持","网络与硬件支持","资产与账号管理"]}],"contact":[{"label":"GitHub","url":"https://github.com/Guuumiho"},{"label":"Agent Code Learning","url":"https://guuumiho.github.io/agent-code-learning"},{"label":"Stroop Brain Train","url":"https://guuumiho.github.io/Stroop-Brain-Train"}],"meta":{"updatedAt":"2026-07-02T22:03:39.950Z"}}