API & system architectureAPI 与系统架构
Designing reliable routes between models, data and products — informed by nearly a decade of work on complex healthcare, supply-chain and enterprise systems.在模型、数据与产品之间设计可靠路径。近十年的医疗、供应链与企业复杂系统经验,是这套能力的底座。
I turn complex systems into products people can actually use — from national healthcare networks and global ERP platforms to AI workflows built and shipped by one person.我把复杂系统变成真正可用的产品——从全国医疗数据网络、跨国 ERP,到一个人也能构建并交付的 AI 工作流。
My work sits where product judgment, system architecture and commercial execution overlap. The interface is only the visible end of a much larger operating system.我的工作位于产品判断、系统架构与商业执行的交叉点。你看到的界面,只是整套运行系统露出水面的部分。
Designing reliable routes between models, data and products — informed by nearly a decade of work on complex healthcare, supply-chain and enterprise systems.在模型、数据与产品之间设计可靠路径。近十年的医疗、供应链与企业复杂系统经验,是这套能力的底座。
Connecting topic discovery, language models, voice and visual production into repeatable creator workflows.把选题、语言模型、语音与视觉生产连接成可重复使用的创作者工作流。
Turning a focused problem into a small, usable product — then putting it in front of real users.把一个明确痛点做成小而可用的产品,并真正交到用户手里。
Choose a chapter instead of reading one long timeline. Each stage added a layer: foundation, ecosystem scale, then independent AI execution.不必从头到尾阅读一堵履历墙。选择一个阶段:基础、复杂生态,再到今天的独立 AI 交付能力。
Designed and built the compute and API routing used by LLM-driven tools, with a focus on stable data flow and structured outputs.从 0 到 1 构建面向 LLM 工具的底层计算与 API 路由,重点解决数据流转稳定性和结构化输出问题。
Explored the complete micro-SaaS loop: identifying a narrow need, building the product and validating an overseas subscription path.跑通微型 SaaS 的完整路径:识别窄需求、构建产品,并验证海外订阅模式。
Led the product architecture of the standardized Metabolic Management Center ecosystem across hardware and software.主导标准化代谢性疾病管理中心(MMC)生态的软硬件产品架构。
Led global ERP, supply-chain and multi-country website systems for cross-border operations.面向跨国业务,主导集团 ERP、供应链及多国官网系统。
Led lifecycle product work for smart thermometers and health applications, connecting device data with mobile healthcare workflows.主导智能体温计与健康 App 的全生命周期产品工作,连接设备数据与移动医疗流程。
Live products and autonomous systems built around real friction — from intelligence discovery to meetings and career decisions.围绕真实阻力构建并持续运行的产品与自动化系统——从情报发现,到会议执行与职业决策。
Radar Station turns scattered global signals into a working intelligence loop. It collects product pain points, AI frontiers and technology shifts, applies an AI judgment layer, then delivers the items worth acting on into Feishu and a reusable opportunity pool.Radar Station 把分散在全球的信号变成一条真正工作的情报闭环:采集产品痛点、AI 前沿与科技变化,经过 AI 判断层,再把值得行动的信息送入飞书和可复用的机会池。
The point is not to collect more links. The point is to make each signal easier to understand, compare and act on.重点不是收集更多链接,而是让每条信号更容易被理解、比较,并转化为行动。
Four independent probes keep each track focused while a shared receiver handles asynchronous processing, AI review, deduplication and storage.四个独立探针保持各轨道的专业性,共享接收器统一完成异步处理、AI 审核、去重与沉淀。
Turn meeting notes or audio into an actionable Eisenhower matrix, ready to copy and export.把会议文字或录音转成可执行的艾森豪威尔四象限待办,支持复制与导出。
Analyze a résumé against a target role and turn vague improvements into specific actions.针对目标岗位分析简历,把模糊的“需要优化”变成具体行动。
A deployed AI-assisted intake workflow for turning messy business requests into structured records a human can review and act on.一个已经部署的 AI 辅助线索工作流,把杂乱的业务需求整理成可供人复核和行动的结构化记录。
The system validates an incoming request, extracts requirements, classifies fit and urgency, detects missing information, stores the result in Supabase/PostgreSQL, and prepares a draft follow-up while keeping the final decision with a human.系统校验输入、提取需求、判断匹配度与紧急程度、识别缺失信息,将结果写入 Supabase/PostgreSQL,并生成跟进草稿,同时保留人工复核和最终决策。
I am interested in practical AI products, workflow architecture and collaborations that lead to something people can genuinely use.我关注实用 AI 产品、工作流架构,以及最终能做出真实可用成果的合作。