Co-founder — Novada LabsNovada Labs 联合创始人

Tong Wu 吴桐

AI automation and agentic systems. Supply chain, logistics, and process excellence. I ship both, end to end. AI 自动化与智能体系统。供应链、物流与流程卓越。两个领域，我都能端到端交付。

Co-founder, Novada Labs. Previously: supply-chain & process excellence at Xiaomi, Amazon, Huawei. Econ/Finance → Supply Chain → AI, in that order. Novada Labs 联合创始人。此前历经小米、亚马逊、华为的供应链与流程管理。从经济金融，到供应链，再到 AI 系统。

Düsseldorf, Germany · EU Blue Card holder — open to select conversations, not actively job-hunting.欧盟蓝卡持有者——欢迎有价值的交流，目前并非主动求职。

Project Management项目管理 AI AutomationAI 自动化 Supply Chain供应链 Logistics物流 Last-Mile Delivery末端配送 Process Excellence流程卓越 Multilingual · CN/EN/DE多语言 · 中/英/德 EU Work Authorization欧盟工作许可 Team Leadership · 100+ FTE团队管理 · 100+ 人 Cross-Border · 4 Countries跨国经历 · 4 国
Email me给我发邮件↗ LinkedIn↗ GitHub↗
310★ GitHub stars · AgentRecallGitHub 星标 · AgentRecall
6,400+ installs / month月安装量
6 published npm / PyPI packages已发布的 npm / PyPI 包
3 languages spoken掌握语言数
4 countries lived / worked in生活/工作过的国家数
Keep going继续往下看
01 — About— 关于

The cross-domain throughline跨领域的那条主线

My path doesn't run in a straight line, and that's the point. Economics & Finance with a computational core, then supply-chain and process-excellence roles at Huawei, Xiaomi and Amazon, now co-founding Novada Labs to ship AI agent infrastructure. Every domain shift kept the same instinct: measure it, fix the root cause, then automate it. 我的路径不是直线，这正是重点。从带有计算内核的经济金融出发，到华为、小米、亚马逊的供应链与流程卓越工作，再到联合创办 Novada Labs、打造 AI Agent 基础设施。每一次跨界都保留了同一个本能：先量化，再找根因，然后自动化。

Read the longer version阅读完整版本（英文）

I hold an M.Sc. in Economics & Finance from the University of Freiburg — though the coursework leaned informatics: computational economics, machine learning, and programming, alongside econometrics and time-series analysis. I also taught and ran behavioral/game-theoretic experiments there as a research assistant. From there I moved into operations: EU compliance work at Huawei, an enterprise KPI-and-delivery platform at Xiaomi (~€110M/yr fulfilment, 99.5% on-time), and process-excellence work at an Amazon delivery station (16 carrier accounts, 100+ KPIs, structured RCA). Today I co-found Novada Labs, where I design and ship the agent infrastructure — memory systems, orchestration frameworks, MCP servers — that other builders use to take AI from proof-of-concept to something that survives contact with a real business process. My first retrieval build was a seminar project at Freiburg — a LangChain + Pinecone RAG assistant (2023) — with no public repo on file.

  • Education教育M.Sc. Econ, Finance & Computing
  • Languages语言CN · EN · DE
  • Now building现在在做Novada Labs
  • Based in常驻Düsseldorf, DE
Domain领域 Where在哪 Focus重点
Econ, Finance & Computing经济金融与计算 Freiburg (M.Sc.) Computational economics, machine learning, programming — the technical seed of everything after.计算经济学、机器学习、编程——后来一切的技术起点。
Process Excellence流程卓越 Huawei → Amazon ISO compliance, then KPI & root-cause analysis at delivery-station scale.从 ISO 合规，到配送站规模的 KPI 与根因分析。
Supply Chain & Logistics供应链与物流 Xiaomi Enterprise KPI & exception-management platform; ~€110M/yr fulfilment at 99.5% on-time.企业级 KPI/异常管理平台；~€110M/年履约额，准时率 99.5%。
AI & Agent SystemsAI 与智能体系统 Novada Labs (now)Novada Labs（现在） AgentRecall, AAM, Plywood, Novada MCP — production, not demos.AgentRecall、AAM、Plywood、Novada MCP——是生产系统，不是演示。
Keep going继续往下看
02 — Building— 在构建

Products shipped at Novada LabsNovada Labs 已发布的产品

Flagship旗舰产品

AgentRecall↗

Persistent memory for AI agents — so corrections actually stick. 为 AI Agent 打造的持久记忆系统——让纠正真正被记住。

310★GitHubGitHub 星标
6,400+installs / mo月安装量
Framework框架

AAM (Agent Autonomous Mode)no public link无公开链接

Orchestration for autonomous agent teams — with a reviewer that isn't the author. 多智能体编排框架——审阅者永远不是执行者本人。

Orchestrator编排者→ Parallel workers并行执行者→ Independent reviewer独立审阅者→ Verifier验证者
Compiler编译器

Plywood Protocolno public link无公开链接

Turns plain-language instructions into executable SOPs — validated, not just claimed. 把自然语言指令编译成可执行的标准作业流程——经过盲测验证，不是自吹。

Blinded RCT盲测 RCTvalidated via验证方式
MCP serverMCP 服务器

Novada MCP↗

40+ tools across 13 platforms — the integration layer AI agents actually need. 40+ 工具、覆盖 13 个平台的 MCP 服务器——AI Agent 真正需要的集成层。

40+tools工具
13platforms平台
v0.9.33on npmnpm 版本
MCP serverMCP 服务器

Novada Proxy MCP↗

Reliable web access for agents, backed by 464 automated tests. 为 Agent 提供稳定的网络访问能力，464 个自动化测试兜底。

464automated tests自动化测试
Show full project detail展开完整项目说明

AgentRecall —AgentRecall — correction-first memory system for AI agents, built on embeddings, vector retrieval, and RRF (reciprocal rank fusion).面向 AI Agent 的纠正优先记忆系统，基于向量嵌入、检索与 RRF（倒数排序融合）。

AAM —AAM — guardrails and human-in-the-loop checkpoints run alongside every step of the orchestrator → workers → reviewer → verifier chain shown above.在上方“编排者→执行者→审阅者→验证者”链条的每一步中，都配有护栏与人工检查点。

Personal / exploratory —个人 / 探索项目 — TchinTalk (栖音), a local-first macOS dictation app exploring on-device speech + AI. Not a Novada product. TchinTalk（栖音），一个探索本地优先听写体验的 macOS 个人项目——不属于 Novada 产品线。

Keep going继续往下看
03 — Professional Experience— 工作经历

Econ/Finance → Supply Chain → AI经济金融 → 供应链 → AI

  1. Mar 2026 — Present至今5 mos个月 Now现在 Full-Time全职

    Co-founder & AI Automation/Infrastructure Engineer联合创始人 & AI 自动化/基础设施工程师

    Novada Labs, Frankfurt / Remote
    • Ships production AI systems:交付生产级 AI 系统： AgentRecall (correction-first agent memory), AAM (multi-agent orchestration — orchestrator → workers → independent reviewer → verifier), and Plywood (NL-to-pseudocode workflow compiler, validated via a blinded RCT).AgentRecall（纠正优先的智能体记忆系统）、AAM（多智能体编排——编排者→执行者→独立审阅者→验证者）、Plywood（自然语言转伪代码工作流编译器，经盲测 RCT 验证）。
    • Integration layer:集成层： built 6 published production packages, including Novada MCP (40+ tools, 13 platforms) and Novada Proxy MCP (464 automated tests); owns TypeScript/Python code, tests, CI/CD and release end to end.发布了 6 个生产级软件包，包括 Novada MCP（40+ 工具，13 个平台）与 Novada Proxy MCP（464 个自动化测试）；端到端负责代码、测试、CI/CD 与发布。
    • Runs the business, not just the code:不只写代码，也管理业务： owns technology-department processes end-to-end — engineering, vendor/procurement, stakeholder communication — as sole technical point of contact for a fast-moving startup.端到端负责技术部门运作——工程、供应商/采购、干系人沟通——是这家快速发展初创公司唯一的技术负责人。
  2. Sep 2025 – Jan 20265 mos个月Full-Time全职

    Process Excellence Manager流程卓越经理

    Amazon, Hamburg (DHH2)
    100+ metrics指标·16 carrier accounts承运商账户·99.5% quality质量·100% safety compliance安全合规
    • In-station operations:站内运营： standardized in-station workflows (receiving, sortation, route planning, loading) across 16 delivery-carrier accounts, via 100+ OTR/UTR metrics.在 16 个配送承运商账户上，标准化了站内作业流程（收货、分拣、路线规划、装车），依托 100+ 项 OTR/UTR 指标。
    • Root-cause analysis:根因分析： led structured RCA (Lean/5S) with corrective-action plans for underperforming carriers — sustaining 99.5% delivery quality and 100% safety compliance.主导结构化根因分析（精益/5S），为表现不佳的承运商制定纠正措施——维持 99.5% 配送质量与 100% 安全合规。
    • Change management:变革管理： directed 100+ L1–L3 associates; built training materials, ran train-the-trainer workshops and monthly audits; shipped an internal AI-assisted reporting tool that saved the team 6+ hrs/week.管理 100+ 名 L1–L3 员工；编写培训材料，开展"培训讲师"工作坊与月度审计；搭建了内部 AI 辅助报表工具，为团队每周节省 6+ 小时。
  3. Jun 2024 – Aug 20251 yr年 3 mos个月Full-Time全职

    Supply Chain Delivery Manager供应链交付经理

    Xiaomi Technology, Düsseldorf
    ~€110M/yr fulfilment/年履约额·99.5% on-time准时率
    • Built the platform:搭建平台： independently scoped requirements for, designed and deployed a company-wide KPI-tracking & exception-management platform — automated dashboards, real-time damage/delay alerts, escalation SOPs, carrier SLA frameworks.独立梳理需求，设计并部署了公司级 KPI 跟踪与异常管理平台——自动化仪表盘、实时破损/延误告警、升级流程、承运商 SLA 框架。
    • Ran it at scale:规模化运营： end-to-end B2B/B2C fulfilment (~320k units/yr, ~€110M) for DE & NL at 99.5% on-time; standardized last-mile 3PL workflows via SAP/CRM/GDP.为德国与荷兰运营端到端 B2B/B2C 履约（约 32 万件/年，约 1.1 亿欧元），准时率 99.5%；通过 SAP/CRM/GDP 标准化末端 3PL 流程。
    • Cross-functional planning:跨部门规划： led CPFR demand planning, S&OP and 12-month rolling supply plans; aligned Sales, GTM and 3PL partners.主导 CPFR 需求计划、S&OP 与 12 个月滚动供应计划；协调销售、市场推广与第三方物流伙伴。
Show full timeline (10 roles)展开完整履历（共 10 段）
  1. Dec 2023 – Jun 20247 mos个月Full-Time全职

    EU Certification / Quality Specialist欧盟认证/质量专员

    Huawei, Düsseldorf

    Managed ISO 9001/14001/45001 compliance across labs, notified bodies and product teams, in three languages.以三种语言管理实验室、公告机构与产品团队之间的 ISO 9001/14001/45001 合规工作。

  2. Oct 2022 – Feb 20241 yr年 5 mos个月

    Graduate Teaching & Research Assistant研究生助教/研究助理

    University of Freiburg

    Taught Finance/Macroeconomics; ran behavioral/game-theoretic experiments at FRIBIS.教授金融/宏观经济学；在 FRIBIS 开展行为与博弈论实验。

  3. Oct 2020 – Nov 2023

    M.Sc. Economics and Finance经济与金融学硕士

    University of Freiburg

    Computational Economics, Machine Learning and programming-heavy coursework (Econometrics, Time-Series Analysis) — more informatics than a typical finance degree. Seminar: LangChain + Pinecone RAG assistant (2023) — his first retrieval build (no public repo on file). Outstanding Student Representative, commencement speaker.计算经济学、机器学习与编程比重很高的课程（计量经济学、时间序列分析）——比典型的金融学位更偏计算机方向。研讨课项目：LangChain + Pinecone RAG 助手（2023）——他的第一个检索系统（无公开代码库）。优秀学生代表，毕业致辞代表。

  4. Jul 2021 – Oct 20214 mos个月Internship实习

    Business Intern商务实习生

    SEG Automotive, China

    Supported procurement processes in SAP; supplier risk management and safety audits in an automotive supply chain.在 SAP 系统中支持采购流程；在汽车供应链中开展供应商风险管理与安全审计。

  5. Nov 2018 – Mar 20201 yr年 5 mos个月Full-Time全职

    Project Management Specialist项目管理专员

    Arshine Pharmaceutical Co., Ltd, Changsha, China

    Managed end-to-end export projects: overseas B2B client development, third-party international logistics, documentation and cross-border customs clearance.管理端到端出口项目：海外 B2B 客户开发、第三方国际物流、单证处理与跨境清关。

  6. Jun 2018 – Sep 20184 mos个月Internship实习

    Business Intern商务实习生

    Harris Teeter, USA

    Supported retail logistics and sales operations in a high-volume US retail environment.在高流量的美国零售环境中支持零售物流与销售运营。

  7. Sep 2014 – Jun 2018

    B.A. Business English (International Trade & Business)商务英语学士（国际贸易与商务）

    Jishou University, China

    Foundation in international trade & economics.国际贸易与经济学基础。

Keep going继续往下看
04 — Skills— 技能

What the numbers above are built on上面这些数字的地基

AI & Agent EngineeringAI 与智能体工程

RAG (embeddings, vector retrieval, RRF)Multi-agent orchestration多智能体编排MCP (server/client)LLM integrationLLM 集成 (Claude API, OpenAI)Evals & benchmarking评测与基准测试Guardrails / quality measurement护栏机制/质量度量PythonTypeScriptPostgreSQLDocker / CI-CD

Process & Operations Excellence流程与运营卓越

LeanSix Sigma六西格玛 (Green Belt)Root-cause analysis根因分析KPI design & monitoringKPI 设计与监控Requirements engineering需求工程Enterprise / key-account management企业/大客户管理ISO 9001 / 14001 / 45001

Certifications认证no verified badge links无认证徽章链接

Six Sigma Green Belt · Certified Business Analysis Professional (CBAP) · Google Data Analytics · Google IT Automation with Python

Languages语言

中文 — Native母语  ·  English — C2  ·  Deutsch — C1

Keep going继续往下看
05 — Off-screen— 屏幕之外

What I read, brew, and practice when I'm not shipping. 不写代码的时候，我在读、在泡茶、在写字。

易经I Ching study— 易经研习
茶道Tea ceremony— 茶道
书法Calligraphy— 书法
徒步Hiking— 徒步
日本語Learning Japanese— 学习日语
Keep going继续往下看