Lin Zhangsheng

Frontend Engineer · Shanghai, China

Experience

Frontend Engineer

HelixLife (Shanghai) Medical Technology Co., Ltd. · Shanghai

  • Led core Web delivery for the Helix AI research collaboration platform (NewIdea writing workstation) from 0 to 1: task creation and management, step-based task flows, SSE real-time logs/stage updates, token usage dashboards, rich-text editing, review, and export — supporting both SaaS and AIO multi-site deployments.
  • Drove frontend architecture and engineering: unified API layer with auth refresh, runtime environment injection, Docker orchestration, Plate/Slate editor pluginization (Diff, Mermaid, tables, serialization), and performance optimization; led code reviews, troubleshooting, and internal knowledge sharing.

Full-Stack Engineer

Frost & Sullivan (Beijing) Consulting Co., Ltd. — Shanghai Branch · Shanghai

  • Participated in product requirements analysis and architecture design, translating complex business needs into scalable technical solutions and shipping multiple projects from concept to production within 3 months.
  • Led frontend technology selection and framework setup, improving page load speed by 35% and achieving 98% design fidelity.
  • Collaborated with backend teams on system modeling and high-availability API development, reducing system failure rate by 45% and maintaining 99.9%+ service stability.
  • Core contributor to a Slate.js-based rich-text editor, improving editing performance by 20% and strengthening product competitiveness.

Software Development Engineer

Shanghai Huaixin Intelligent Technology Co., Ltd. · Shanghai

  • Developed Web-based frontend pages for the MES system.
  • Performed secondary development on Siemens OpCenter EXPR based on business requirements.
  • Owned module-level development, debugging, and maintenance.

Frontend Engineer

Shanghai Lingyun Information Technology Co., Ltd. · Shanghai

  • Built ToB admin systems using Vue 3, TypeScript, Element Plus, and ECharts for complex data visualization and permission configuration.
  • Implemented features from prototypes and resolved complex permission configuration challenges.

Skills

Frontend Architecture

  • JavaScript / TypeScript: ES6+, async/event loop, modularity, engineering practices; deep understanding of scope, closures, prototypes, design patterns, and maintainable code.
  • React ecosystem: Hooks, Context, component abstraction, state management (Zustand, etc.), routing, forms, permissions, micro-frontends; performance optimization (memo, useMemo, useCallback, profiling, code splitting, lazy loading).
  • Vue ecosystem: Vue 3 Composition API, Pinia, Vue Router; component library design and complex module decomposition; reactivity and rendering internals.
  • Build & optimization: Vite/Webpack, Babel, PostCSS; multi-environment CI pipelines; first paint, caching, on-demand loading, bundle analysis, CDN, error monitoring, and analytics.

Full-Stack & Cross-Platform

  • Node.js full-stack: TypeScript server-side engineering with NestJS/Hono — API design, validation, auth, permissions, error handling, and logging.
  • Full-stack frameworks: Next.js, TanStack Start, Nuxt — CSR/SSR/SSG/ISR, data fetching/caching, Server Actions, and middleware.
  • Databases: PostgreSQL schema/index design, query optimization, transactions; Drizzle ORM type-safe modeling, migrations, and data access layers.
  • Cross-platform: React Native & Expo engineering and release; familiarity with Taro mini-programs and Tauri desktop apps.

AI

  • AI Agent coding: proficient with Cursor, Codex, Claude Code — task decomposition, multi-turn implementation, prompt/rules/context management, and reusable skills for scaffolding, components, APIs, logging, and performance analysis.
  • Agent development: end-to-end intelligent agent apps with Mastra, Vercel AI SDK, LangChain — model integration, tool/function calling, RAG, structured output, streaming, evaluation, and observability.
  • Python: self-taught; data processing and automation scripting.
  • AI workflows: hands-on with Dify, n8n, Inngest for data ingestion, triggers, orchestration, persistence, and notifications.

Education

Bachelor of Engineering · Electronic Information Engineering

Shanghai University of Electric Power

Projects

AI-powered research and writing workstation: create and manage AI writing tasks, view real-time execution logs and dashboards, edit/review/export in a rich-text environment; supports SaaS and on-premise multi-site deployments.

Next.js 14 (App Router), React 18, TypeScript, Tailwind CSS, Zustand/Immer, umi-request, SSE, Plate (Slate) editor, Chart.js, Zod/TanStack Form, pnpm, biome/Husky.

  • Built core business pages: task list/pagination, task detail step flows, dashboards, and token usage with optimized data fetching, state sync, and interaction (visibility-based refresh, debouncing).
  • Designed unified API layer: umi-request prefix, Bearer auth, 401/token refresh via @helix/helix-token-manager, business error routing, and i18n error messages.
  • Implemented streaming AI output and task logs via SSE, with Streamdown + Markdown for incremental rendering and Safari compatibility.
  • Complex multi-type task forms with field dependencies, validation hooks, and conditional rendering.
  • Plate plugin system for Diff, Mermaid, tables, and serialization — supporting long-document editing, comparison, and export.

Vertical AI collaboration platform for medical and research workflows. Integrates Dify orchestration and CopilotKit tool-calling to expose page capabilities to AI assistants — covering literature analysis, academic writing, and grant proposal assistance.

Next.js, React, Tailwind CSS, Dify, Docker, CopilotKit

  • Maintained and extended writing workbench pages, reusing one codebase across a dozen deployed writing scenarios.
  • Architected "AI Agent + immersive workbench" integration, shortening core writing cycles from weeks to hours.
  • Customized Platejs research editor with AI text selection, revision mode, and smart reminders; solved atomic incremental rendering for conflict-free AI/human content merging, improving grant proposal output efficiency by 70%.

Enterprise intelligent writing editor built on React and SlateJS. Monorepo architecture with pnpm workspaces for modular, extensible document editing.

React, TypeScript, SlateJS, Vite, pnpm

  • Extended SlateJS with configurable core editor and plugin architecture; optimized large-document editing performance.
  • Built collaborative editing with Slate.js + Yjs and Hocuspocus for real-time sync and persistence.
  • Developed term data processing, multi-format export (Word, plain text), and ECharts chart generation.
  • TypeScript type-safe development with comprehensive unit tests and modular code organization.

Languages

  • Mandarin: Native
  • English: Proficient