From AI tool user to workflow builder
The current goal is to move beyond tool fluency and learn how to design repeatable systems that produce useful outputs.
AI Portfolio / Personal Positioning System
我正在用 AI 构建项目、流程、分析系统、内容系统和个人知识系统。
AI product operations candidate with frontline 3C and smart-hardware experience, using AI to connect user needs, content production, data feedback, workflow design, and public project evidence.
Personal Capability Narrative
A candidate for AI product operations, AI product assistant, content operations, and data analysis roles. Combines frontline 3C and smart-hardware business experience with long-term AI tool usage, user communication, content production, data feedback analysis, and workflow building.
The current goal is to move beyond tool fluency and learn how to design repeatable systems that produce useful outputs.
I care about how AI connects with data, content, operations, product decisions, and human review rather than existing as a separate toy.
I prefer learning through projects, dashboards, workflows, and public artifacts so that ability can be inspected and improved.
My advantage is not pretending to know everything, but learning quickly, shipping prototypes, and tightening them with feedback.
AI Usage Experience
The timeline shows how AI moved from writing assistance into content production, prompt design, data understanding, requirement breakdown, and project iteration.
Began using ChatGPT for content understanding, writing assistance, prompt exploration, and task decomposition.
Experienced mainstream domestic AI tools during the rapid growth of AI software, with high-frequency use of DeepSeek and Tongyi Qianwen.
Uses OpenAI GPT and Codex as daily working tools for content systems, project iteration, requirement breakdown, data interpretation, and automation-oriented workflows.
Capability Matrix
Each capability is connected to project proof, current stage, and the keywords that describe how it is being built.
Use GPT, Codex, DeepSeek, Tongyi Qianwen, Gemini, ComfyUI, and Jimeng to design repeatable input-processing-review-output systems.
Turn ambiguous goals into structured prompts, evaluation criteria, output templates, and review loops.
Use AI coding agents to scaffold projects, inspect files, implement changes, and verify build results.
Collect, clean, segment, visualize, and explain data with an operations and business perspective.
Build the habit of defining metrics, dimensions, comparison windows, and decision-relevant cuts.
Translate analysis into interfaces that help readers compare, scan, and decide.
Connect user feedback, product experience, workflow constraints, business goals, and operational next steps.
Convert source material into localized, platform-ready, reusable content packages for articles, scripts, visual prompts, and voiceover drafts.
Collect fragmented information, organize it into themes, and communicate the implications clearly.
Turn continuous learning into a structured path with evidence, outputs, and compounding artifacts.
Role Fit Logic
I have a foundation in frontline business, user communication, and smart-hardware product understanding, while continuously using AI tools to make work more structured and repeatable. Compared with a single execution role, I am better suited to work that connects requirements, content efficiency, data feedback, and project progress in real business scenarios.
From ChatGPT to Codex, DeepSeek, Tongyi Qianwen, Gemini, ComfyUI, and Jimeng, AI has been used in real tasks rather than only casual experiments.
3C and smart-hardware retail experience supports user communication, product explanation, sales conversion, store operations, and product training.
Current projects show content workflows, data dashboards, user-feedback analysis, and operation review thinking through public artifacts.
Featured Projects
Projects are structured as case studies: value, process, outputs, business meaning, learning, and next steps.
3C 门店经营 AI Agent:面向智能硬件单店经营的 AI 助手
一个面向 3C / 智能硬件单店经营场景的 AI PM 作品集 MVP,用本地样例数据和 Rules / Mock Agent 串联经营看板、问题诊断、知识库、话术、日报和质检闭环。
把体检报告转化为可理解、可执行、可追踪的健康行动。
一个将体检报告转化为健康行动的 AI 产品设计项目,面向普通体检用户和家庭健康管理者,覆盖报告上传、OCR 识别、指标结构化、异常项解释、关注优先级、健康行动计划、AI 追问、安全审核和长期健康档案。
帮助用户从难懂的体检指标进入更清晰的健康信息理解、复查提醒、医生沟通准备和长期健康档案管理,同时明确不替代医生判断。
从进度条 UI 出发,构建可复用的任务状态反馈系统。
围绕上传、导出、AI生成、视频转码、审核等耗时任务场景,系统拆解用户在等待过程中的不确定性、焦虑感、重复操作和中途退出问题,将“加载中”升级为可感知、可解释、可恢复、可复用的任务状态反馈机制。
通过统一耗时任务反馈,降低用户等待不确定性、重复点击和中途退出风险,同时减少不同模块重复开发和体验割裂。
From Steam reviews to operating insight.
A data project around Steam game review collection, cleaning, analysis, visualization, and game-operation recommendations.
Shows how player feedback can be turned into practical signals for product operation, user issue discovery, localization, and content decisions.
Evidence-bound Game Business Analytics Dashboard
基于公开 Steam 信号、公开销量节点、媒体评分、竞品对比和有限评论样本,整理为证据边界版游戏商业分析看板。项目展示公开数据组织、结论边界管理、Dashboard 结构设计、响应式 QA 和 Cloudflare Preview 部署验证能力。
把现象级游戏项目拆成公开信号、数据看板、证据边界和作品集叙事,体现面向 AI 产品与数据运营岗位的结构化分析能力。
A full-chain product and business research case.
A structured research project around Apple Watch product positioning, user needs, feature evolution, ecosystem value, business model, supply chain, and competitive landscape.
Trains the ability to connect product design, user scenarios, ecosystem strategy, commercial value, and structured report writing.
English AI newsletter to Chinese multi-platform content pack.
A workflow for understanding deeplearning.ai Andrew's Letter and transforming it into Chinese articles, scripts, image prompts, voiceover text, and platform assets.
Compresses the AI content production cycle while keeping human review as the quality gate and adapting one source into multi-platform assets.
Image resume to editable, reviewed, exportable PDF.
A document automation workflow connecting OCR, structured extraction, human review, layout generation, and PDF output.
Reduces manual document formatting work while preserving accuracy through review checkpoints.
一个围绕 AI 产品运营 / 产品助理方向搭建的可持续迭代个人作品集系统。
将个人定位、产品项目、产品分析、研究思考、文档资产、SEO、审计检查和 Cloudflare 部署流程整合成一个可公开访问、可持续更新的内容产品系统。
让个人能力、项目资产、产品分析和长期思考不再分散在简历或文件夹里,而是沉淀为一个可访问、可维护、可审计、可持续更新的内容产品。
A long-term knowledge base for reading, research, and AI-assisted thinking.
A personal system for recording books, notes, ideas, AI-assisted summaries, and essay drafts as long-term cognitive assets.
Builds a durable learning asset that supports communication, research, and product thinking.
Career Evidence
These cards keep the resume evidence compact and connect it back to user communication, product explanation, training, activity execution, and data feedback.
Proof of Work
This layer separates confirmed proof from placeholders, so reviewers can quickly see what is already real.
The portfolio is deployed on Cloudflare Pages and can be reviewed as a public static site.
Open proofThe source is version-controlled and prepared for a GitHub-based Cloudflare Pages workflow.
Open proofThe production site is hosted on Cloudflare Pages with static build output and public SEO files.
Open proofPortfolio content is managed from structured data and rendered into static pages.
readyFive project case-study pages and six thinking draft pages are generated from data.
Open proofThe build outputs an audit report that tracks placeholders and next proof gaps.
Open proofAI Workflow System
The workflow area is the core of this portfolio: it shows how AI is used as a system layer, not a one-off chat interface.
Check whether the metric story matches real player context before writing conclusions.
Dashboard, insight memo, risk keywords, operation recommendations
New questions from the dashboard become the next analysis cut.
Review accuracy, tone, examples, and platform fit before publishing.
WeChat article, short-video script, prompt pack, voiceover text
Audience response and editorial notes improve the next weekly template.
Verify names, dates, roles, and achievements before final rendering.
Editable structured data and exportable PDF resume
Corrections update parsing rules and layout constraints.
Learning Timeline
The path is designed to compound: each stage creates stronger evidence for the next stage.
Started using AI as a daily assistant for search, writing, translation, and problem decomposition.
Built the habit of reading English AI material and turning it into understandable notes.
Shifted from vague analysis to metric definitions, dimensions, and comparison thinking.
Began using Python and Pandas concepts for data cleaning, grouping, and analysis.
Started turning analysis into interfaces and stories readers can scan.
Used AI coding workflow to inspect requirements, edit files, build, and verify outputs.
Connected source understanding, localization, scriptwriting, visual prompts, and publication assets.
Started treating reading and reflection as reusable long-term assets.
Connected projects to roles, business value, user feedback, and industry scenarios.
Organized learning, projects, and positioning into a site that can be deployed and updated.
Thinking / Essays
Thinking about AI as a layer around work rather than a single app.
A reflection on why AI matters most when it is embedded into repeated processes.
From access to information toward execution and synthesis.
Why active learning and output ability become more important when information is abundant.
A practical framework for AI-era project building.
A simple model for connecting technical ability to business outcomes.
The shift from trying tools to designing systems.
A note on why the next level of AI ability is workflow design.
Why motivation to keep learning becomes a durable advantage.
A personal argument for curiosity, execution, and visible learning artifacts.
Turning reading, notes, and reflection into long-term cognitive assets.
A plan for capturing inputs, summarizing ideas, linking themes, and producing essays.
从多个 Codex 项目 prompt 复盘中提炼出的 AI 工程协作提示词方法论。
通过复盘 RetailOps Agent、黑神话悟空分析、Andrew's Letter 内容流水线和个人主页项目,提炼适合工程项目的 Codex prompt 方法论。
Education
Bachelor / Energy and Power Engineering
Resume / Profile
Connecting AI tool usage, 3C frontline experience, content production, and data feedback into AI product operations work.
Final resume PDF has been added. Phone and WeChat remain private and are not displayed on public pages.
Download ResumeContact
Email is public and ready. Phone and WeChat are kept private for the public portfolio.