个人简介

拥有 5 年工程研发经验，近年主要专注于 AI Agent、LLM 应用和自动化工作流系统的设计与落地，熟悉从 Prompt 设计、工具调用、记忆管理到评测闭环的完整链路。

擅长将复杂业务流程抽象为可观测、可重试、可扩展的 Agent 系统，具备从 0 到 1 交付 AI Copilot、多代理协作工具和企业自动化助手的经验。

个人技能

• Agent / LLM：Prompt Engineering、Function Calling、RAG、长期/短期记忆、规划执行、Multi-Agent Orchestration
• 框架与语言：TypeScript、Python、Node.js、React、Next.js
• 工程化：Docker、CI/CD、异步任务队列、日志追踪、评测脚本、灰度发布
• 工具生态：OpenAI / Anthropic API、MCP、向量检索、Webhook、Workflow Automation

工作经历

智序实验室 · AI Agent 工程师2023.03 - 至今 · 杭州

• 负责企业级 AI Agent 平台研发，主导 Agent Runtime、工具注册、会话记忆与任务编排模块设计
• 设计多代理协作流程，将需求分析、检索、执行、校验拆分为独立角色，提升复杂任务完成率
• 建立日志追踪与评测机制，围绕调用成功率、任务耗时和答案质量持续优化系统表现

星帆智能 · LLM 应用开发工程师2021.07 - 2023.02 · 上海

• 参与 AI Copilot 产品研发，构建面向运营和知识工作的问答、总结、生成与审批辅助能力
• 基于 TypeScript + Node.js 封装模型网关与工具调用层，统一接入检索、表单、知识库与第三方 API
• 推动 Prompt 模板化和回归评测流程，减少版本迭代中的输出漂移问题

云迹工作室 · AI 自动化开发实习生2020.06 - 2021.06 · 南京

• 参与内部自动化助手原型开发，支持邮件整理、知识检索和任务分发等场景
• 编写数据清洗与流程脚本，协助搭建知识入库与 FAQ 检索链路
• 整理 Agent 实验记录和效果对比文档，支持后续迭代决策

重点项目经历

企业知识助手 Agent 平台2024 - 至今

主要技术栈：TypeScript + Node.js + MCP + Vector Search

• 从 0 到 1 搭建企业知识助手平台，支持问答、文档总结、流程指引和工单协助等场景
• 抽象统一工具协议，接入搜索、数据库、Webhook 和内部业务系统，提升 Agent 能力扩展效率
• 设计会话记忆与检索增强链路，降低幻觉并提升多轮任务连续性

多代理任务编排系统2023 - 2024

主要技术栈：Python + Workflow Engine + Redis + Postgres

• 将复杂任务拆解为 Planner、Researcher、Executor、Reviewer 等多个 Agent 协同执行
• 支持失败重试、人工接管、状态恢复和步骤级审计，适用于长链路自动化任务
• 引入基准任务集与自动评测脚本，用于比较不同模型和 Prompt 策略的效果差异

开源 AI Agent 模板仓库2026

主要技术栈：Markdown + JavaScript + Cloudflare Pages

• 将 AI Agent 方向的个人作品整理为可公开复用的开源模板仓库
• 提供中英文双语示例、部署说明和面向 Agent 开发者的履历表达方式
• 将示例内容全部替换为虚构资料，降低开源时的隐私暴露风险

开源项目

agent-workflow-starter · AI Agent 工作流模板260+ Stars

• 一个面向 AI Agent 项目的起步模板，包含工具调用、任务状态和日志追踪的基础结构
• 适合作为 Agent Demo、内部助手或自动化工作流项目的脚手架

prompt-eval-lab · Prompt 与 Agent 评测工具120+ Stars

• 用于管理基准数据集、比较多模型输出并追踪 Prompt 版本效果
• 支持批量运行、结果导出和简单的回归评测流程

教育经历

Summary

Engineer with 5 years of product and platform experience, recently focused on AI Agents, LLM applications, and workflow automation. Familiar with the full stack of prompt design, tool use, memory, retrieval, and evaluation loops.

Experienced in turning complex business processes into observable, retryable, and extensible agent systems. Has shipped AI copilots, multi-agent workflows, and internal automation assistants from scratch.

Skills

• Agent / LLM: Prompt engineering, function calling, RAG, short-term and long-term memory, planning, execution, multi-agent orchestration
• Frameworks & Languages: TypeScript, Python, Node.js, React, Next.js
• Engineering: Docker, CI/CD, async job systems, tracing, evaluation scripts, staged rollout
• Tooling: OpenAI / Anthropic APIs, MCP, vector retrieval, webhooks, workflow automation

Work Experience

Sequence Intelligence Lab · AI Agent EngineerMar 2023 - Present · Hangzhou

• Build an enterprise AI Agent platform, leading the design of agent runtime, tool registration, conversational memory, and task orchestration modules
• Designed multi-agent collaboration flows that split complex tasks into analysis, retrieval, execution, and review stages
• Established tracing and evaluation loops around success rate, latency, and answer quality to improve system reliability

Star Sail Intelligence · LLM Application EngineerJul 2021 - Feb 2023 · Shanghai

• Worked on an AI copilot product for operations and knowledge work, covering Q&A, summarization, content generation, and approval assistance
• Built a unified model gateway and tool-calling layer with TypeScript + Node.js, integrating retrieval, forms, knowledge bases, and third-party APIs
• Introduced templated prompts and regression-style evaluation workflows to reduce output drift between releases

Cloudtrail Studio · AI Automation InternJun 2020 - Jun 2021 · Nanjing

• Prototyped internal automation assistants for email triage, knowledge lookup, and task routing
• Wrote data-cleaning and process scripts to support ingestion and FAQ retrieval pipelines
• Maintained experiment notes and comparison documents for agent iteration

Key Projects

Enterprise Knowledge Assistant Platform2024 - Present

Tech stack: TypeScript + Node.js + MCP + Vector Search

• Built an enterprise knowledge assistant platform from scratch for Q&A, document summarization, process guidance, and ticket support
• Abstracted a unified tool protocol for search, databases, webhooks, and internal business systems
• Designed memory and retrieval-enhanced flows to reduce hallucination and improve multi-turn continuity

Multi-Agent Task Orchestration System2023 - 2024

Tech stack: Python + Workflow Engine + Redis + Postgres

• Broke complex tasks into Planner, Researcher, Executor, and Reviewer agents working together
• Supported retries, human handoff, state recovery, and step-level audit logs for long-running automation
• Built benchmark tasks and automated evaluation scripts to compare models and prompt strategies

Open AI Agent Resume Template2026

Tech stack: Markdown + JavaScript + Cloudflare Pages

• Packaged AI Agent focused portfolio content into a reusable public template repository
• Added bilingual examples, deployment instructions, and resume phrasing tailored for agent engineers
• Replaced all sample data with fictional information to make open sourcing safer

Open Source

agent-workflow-starter · AI Agent Workflow Starter260+ Stars

• Starter template for AI Agent projects with tool calling, task state handling, and tracing basics
• Suitable for demos, internal assistants, and automation workflow prototypes

prompt-eval-lab · Prompt and Agent Evaluation Toolkit120+ Stars

• Helps manage benchmark datasets, compare multi-model outputs, and track prompt version quality
• Supports batch runs, exportable results, and lightweight regression-style evaluation

Education