AI Product · Forward-Deployed Engineer (FDE)AI 产品 · 交付工程师 (FDE)

Frank (Aoxuan) Wang · 王奥暄王奥暄 · Frank (Aoxuan) Wang

In an age of technology surplus, the real subject of enterprise AI is the messy here and now. 技术溢出的时代,
企业 AI 的课题,
在每个当下的琐碎现场。

Silicon Valley talks agents, models, and workflows. But walk into a real business and the staff are still on WeChat, Excel, and paper — keying in, cross-checking, and judging every case by hand. The real opportunity isn't flashier tech; it's bringing proven AI into these messy, on-the-ground workflows and making it actually run.

I'm Frank (Aoxuan) Wang, an AI Product Manager / Forward-Deployed Engineer who starts from people and workflows. Below are three — a health store, an after-school center, a moving crew. Each one is a live demo you can click through.

硅谷在聊 Agent、模型、工作流;可我走进真实的企业现场,员工还在用微信、Excel、纸质单据,靠人工一条条录入、核对、判断。真正的机会不在更炫的技术,而在把已经成熟的 AI 带进这些琐碎现场,让它真的跑起来。

我是王奥暄(Frank),一名从人和工作流出发的 AI 产品经理 / 交付工程师(FDE)。下面是三个现场——健康门店、课后教育中心、搬家仓储团队。每一个都是可以点开体验的真实交付。
See the three demos看三个交付现场 About me认识我

Work · Three Live Demos作品 · 三个交付现场

A business never buys "new tech" —
it buys a problem solved. 企业买的从来不是技术的新旧,
而是问题有没有被解决。

Three businesses, three different messes. Same method: find where the staff get stuck, judge which step is worth AI, ship something they can verify. 三家企业,三种不同的乱摊子。方法只有一个:找到员工每天卡住的地方,判断哪一步值得用 AI,然后交付一个他们能亲手验证的东西。

①AI Health Platform · ChengduAI 健康管理中心 · 成都

Interactive Demo Available交互 Demo · 可点击体验

Full-Journey Store AI & Scoring Algorithm全流程门店 AI 与自研评分算法

Unified multimodal data (questionnaires, diagnostics, OCR reports) with a custom 10-dimension AI scoring algorithm. Engineered an end-to-end customer journey: intake, assessment, radar chart, routing, and AI advisory.打通问卷、检测与体检报告，基于规则 + 大模型自研 10 维健康评估算法。重塑从“建档—测评—雷达图—干预方案—AI 顾问”全流程体验。

ImpactReduced service time from 20 to <5 mins; cut review workload by 90%; live in a real store.成效服务时长 20 分钟 → <5 分钟，病例审阅量降 90%，落地真实门店。

Stack: Experience Design · Core Scoring · Multimodal Data定位：体验设计 · 核心算法 · 多模态集成Note: Demo is anonymized — local mockup data, production-grade frontend logic.注：Demo 已脱敏，基于本地数据复刻生产环境前端逻辑。

②ICEC Center · Millbrae, CAICEC 课后教育中心 · 加州 Millbrae

Website · Live Preview网站 · 静态预览

Conversion-First Web & GEO / AI-Search Funnel转化优先的 GEO + AI 搜索获客门户

Rebuilt the digital presence around high conversion: enabled instant program comprehension, real-time lead routing to owners, and GEO & AI-search optimization for local query capture.解决优质课程线上“隐形”痛点，围绕“高转化”重建网站：家长极速理解价值、一键推送表单至店主，结合 GEO 区域定位与 AI 搜索优化精准捕获高意向流量。

ImpactAccelerated client trust, boosted real-time lead push and AI search visibility.成效建立极速信任，提升线索实时转化与 AI 搜索可见度。

Stack: Conversion Architecture · Hyper-Local SEO · AI Search定位：转化 Web 架构 · GEO 搜索引擎优化

icec-afterschool.com
ICEC After School Program — conversion-first homepage (hero, trust badges, pickup, programs, enroll form)

Homepage — static preview. Five screens, one job: answer a parent's five questions and end at the enroll form.首页 · 静态预览。五屏只做一件事:依次回答家长的五个问题,收尾落在报名表单。

③FSS Storage · ChicagoFSS 搬家与仓储 · 芝加哥

Mobile Tool · 3 Static Views手机工具 · 3 页静态图

Zero-Learning-Curve Field Operations Tool0 学习成本的现场极简作业工具

Replaced manual notes for solo operators + gig workers. Built a mobile-first, zero-login app with job scheduling, truck load metrics, tool checklists, and photo-annotated warehouse bins.解决单人带零工全靠脑力调度、易丢物品的痛点。打造移动优先、免登录工具：集成订单调度、卡车装载率 / 工具清单、带照片备注的动态仓库储位——“一键查物”。

ImpactZero learning curve for gig staff; eliminated inventory loss and scheduling friction.成效零工 0 门槛上手，大幅降低现场误操作与库存混乱。

Stack: Field Distillation · Mobile Web · Automation定位：现场需求提炼 · 移动端 Web 开发

FSS field-ops tool — schedule view with calendar and today's tasks FSS 现场作业工具 — 日程视图
1 · Schedule — today's tasks + month at a glance1 · 日程管理 — 当天待办 + 月历总览
FSS field-ops tool — project list with status, crew, trucks and progress FSS 现场作业工具 — 项目列表
2 · Projects — status, crew, trucks, checklist progress2 · 项目管理 — 状态、人员、车辆、清单进度
FSS field-ops tool — new project form: tap-to-pick job type, rooms, crew, trucks FSS 现场作业工具 — 新建项目表单
3 · New project — all tap-to-pick, nothing to type3 · 添加项目 — 全部点选,不用打字

Other Work其他项目

Same thread:
turning data & AI into systems that ship. 同一条主线:
把数据与 AI 变成能落地的系统。

AI Product Analyst · JustCare · 2025–2026AI 产品分析师 · JustCare(加十康) · 2025–2026

Smart Supplement Recommendation Engine智能补剂推荐引擎

Built a Python dynamic goal-driven routing engine that customizes patient-intake flows in real time; integrated Google/Qwen APIs to parse unstructured medical records (OCR) into structured 6-dimension health radars — ~90% less manual review; designed an AI subscription health-manager agent on WeChat for lifecycle tracking and renewals.用 Python 构建目标驱动的动态问诊路由引擎,实时定制患者问诊流程;接入 Google / Qwen API 把非结构化医疗报告 OCR 解析为结构化 6 维健康雷达,人工审阅时间减少约 90%;并在微信生态设计了 AI 订阅制健康管家智能体,用于生命周期追踪与续费提升。

Python · LLM API · OCR · E-commerce recommendation logic电商推荐逻辑

Personal Full-Stack MVP · Live个人全栈 MVP · 线上运行中

Location-Based Generative Content Recommender基于地理位置的生成式内容推荐

Solo end-to-end: needs analysis → API → front-end UI. Integrated an LLM API with geolocation to build a real-time, location-based generative content recommender; containerized and deployed (DevOps).单人端到端:需求分析 → API → 前端 UI。整合 LLM API 与地理定位,构建实时的、基于位置的生成式内容推荐应用;容器化并完成部署(DevOps)。

Full-stack · LLM · Containerized deploy全栈 · LLM · 容器化部署

Visit the live site访问线上版本

About关于我

Frank (Aoxuan) Wang 王奥暄王奥暄 Frank (Aoxuan) Wang

I turn manual, fragmented workflows into purpose-fit systems that ship — product, algorithms, and last-mile deployment, delivered end to end by one person. On a lean team I'm the utility player: from product strategy down to fixing dead links and QA. That last mile is what FDE means to me. 我把手工、割裂的流程,重构成能落地的专用系统 —— 产品、算法、到最后一公里部署,一个人端到端交付。在精益团队里我是那个"全能替补":从产品策略,一直做到修死链接和 QA。这最后一公里,就是我理解的 FDE。

Beyond work工作之外

  • Jazz drummer & band founder. I use the small-ensemble analogy for work: nowhere to hide, listen intensely, improvise and cover for teammates.爵士鼓手、乐队创始人。我用"小编制乐队"类比工作:无处可藏、专注聆听、即兴补位。
  • Founder & lead organizer of a 700-member cultural community, running 5+ large-scale music events a year.创办并主理一个 700 人规模的文化社群,每年组织 5 场以上大型音乐活动。

Skills技能

AI & Agents

Agent workflows智能体工作流Prompt engineering提示词工程LLM API integrationLLM API 集成OpenClaw

Build构建

Vue 3 / TypeScriptPythonClaude Code

Data & Product数据 & 产品

SQLRGTM strategyGTM 策略A/B testingA/B 测试

Experience经历

  1. 03/2026 – Present · Shanghai / Chengdu2026.03 – 今 · 上海 / 成都

    Assistant to the President & AI Transformation Lead · XinChao Media Group总裁助理 & AI 转型负责人 · 新潮传媒集团

    Leading a nationwide AI transformation: agentic workflows across 5 departments (~1,000 employees), turning monthly/quarterly reports into daily automated outputs; driving core initiatives for the AI healthcare subsidiary.主导全国性 AI 转型:5 个部门约 1000 人的智能体工作流,月度/季度经营报告变每日自动产出;推进 AI 健康子公司核心项目。

  2. 09/2025 – 03/2026 · Shanghai (Remote)2025.09 – 2026.03 · 上海(远程)

    AI Product Analyst · JustCareAI 产品分析师 · JustCare(加十康)

    Python goal-driven intake engine; OCR pipeline cutting manual review by ~90%.Python 目标驱动问诊引擎;OCR 解析链路,人工审阅减少约 90%。

  3. 09/2024 – 12/2024 · Chicago2024.09 – 2024.12 · 芝加哥

    UI Policy Consultant · UChicago Policy Lab × IDESUI 政策咨询顾问 · UChicago Policy Lab × IDES

  4. 01/2023 – 06/2023 · Seattle2023.01 – 2023.06 · 西雅图

    Data Analyst · WA Education Reform Study数据分析师 · 华盛顿州教育改革研究

Education教育

  • University of Chicago · Master of Public Policy (MPP) · 2025芝加哥大学 · 公共政策硕士 (MPP) · 2025
  • University of Washington · BS Economics & BA Psychology · 2023华盛顿大学 · 经济学理学士 & 心理学文学士 · 2023