I started in front-end development and have spent 8 years delivering products and full-stack solutions. I am currently a Full-stack Developer / Product at MarsLadder, responsible for student-facing product flows, front-end implementation, and service collaboration in online education. I previously worked at ThoughtWorks on complex business systems, mobile applications, and consulting engagements.

I also build AI applications and local development tooling around Codex, Claude Code, and Hermes, including Skills, Memory, MCP, and reusable unattended AI workflows, while retaining hands-on product, front-end, and back-end delivery capability.

• Student products and data: Daily Practice, Streak / Shield, MarsCoin / Exp, Resource Shop, learning outcomes and feedback, Redash
• Front-end engineering: TypeScript, React, React Router, Vue, Vite, Webpack, GraphQL / Apollo, Zustand, styled-components, KaTeX / LaTeX
• Back-end / Ops: Node.js, Apollo GraphQL, SSE, Express, AWS, Java / Kotlin, MySQL, Flyway, MongoDB, Redis
• AI and development tooling: Codex, Claude Code, Hermes, Skills, Memory, MCP, Python CLI, Prompt Workflow
• Engineering quality: Jest, MSW, Playwright, TDD, CI/CD, GitHub Actions, Docker, Nginx, Shell, NewRelic, PagerDuty
• Mobile and games: React Native, iOS app builds, App Store releases, Cocos, Unity

2023.9 - Present

• Built student-facing core flows for task details, practice, marking, results, and learning reports with React, TypeScript, GraphQL, and Zustand; handled routing, state management, and question-count rules.
• Advanced the Daily Practice gamification loop: after practice, students see Streaks, milestone celebrations, and next-day encouragement; missed practice is protected by Shield with a clear next step.
• Advanced the first version of the Resource and Shield Shop, connecting practice rewards and result-page presentation with MarsCoin balance, Shield inventory, and purchases into an auditable student resource loop.
• Maintained the Realus and MarsLadder websites, advancing SEO and landing-page optimization; tracked traffic with Redash, with organic traffic trending upward recently.
• Owned the Node.js BFF / Apollo GraphQL layer for course, question, practice, and user context; completed front-end integration, batched queries, and cross-service coordination.
• Led production releases for the reward-resource-service / reward-resource-db-migration AWS microservices; currently collaborating on the economy AWS microservice refactor and clarifying Flyway database migration flows.
• Built Redash operational reports for gamification features, using practice, rewards, levels, and course data to validate feature usage.

• Provided a unified AI-tool entry point for students and teachers, including access control, content rendering, and conversation capabilities.
• Integrated AI Tutor into question-answering and practice scenarios, handling question context, course state, and conversation recovery.
• Built the ariadne AWS microservice map tool, statically extracting GraphQL, REST, Kafka, and front-end query dependencies. It exposes CLI and MCP queries for AI systems to retrieve service relationships on demand, reducing repeated code-scanning context and improving cross-service diagnosis.
• Governed Skill context budgets in skill-dreamer through scoped availability, on-demand resolution, and description-budget checks. This avoids unrelated preloading from consuming context so AI can understand tasks and execute consistently in an automated harness.

• Built front-end state and error-recovery mechanisms covering Vite chunk preloading, React Router error boundaries, bootstrap reloads, and Nginx cache control to reduce post-release blank-screen risk.
• Applied Jest, MSW, unit tests, and E2E testing to core features; addressed loading behavior, long-content performance, and post-release issues to improve production stability.

2020.12 - 2023.8

• OPPO after-sales settlement system: designed a Vue form DSL covering layout, validation, linkage, tables, and imports so complex business rules could be completed through a clear interface.
• Resolved duplicate approval-information issues for OPPO by clarifying root causes with business, QA, PM, platform-service, and BPM teams; used process diagrams to explain the solution, supported regression testing, and closed the loop.
• Daimler sales-terminal React Native upgrade: completed SSO, routing, camera, video, device notifications, and legacy-plugin compatibility across iOS and Android scenarios.
• Huawei low-code engine: contributed to the canvas, drag-and-drop, layout, component library, event management, plugin mechanisms, undo / redo, and cross-technology component integration.
• PingCAP website rebuild / SEO optimization: built a static content site with Gatsby, React, GraphQL, and a CMS, supporting Markdown rich text, table-of-contents generation, code highlighting, and mobile adaptation.
• REA PropTrack Australian property valuation system: contributed to Node.js back-end work, testing, and workflow maintenance with Express, MySQL, Redis, AWS, and Docker.
• Organized sharing sessions in the ThoughtWorks game-engine technical community; prepared reference materials, game demos, and recordings, and continuously collected feedback to help colleagues from different technical backgrounds understand game technology.

2018.9 - 2020.11

• Contributed to the client, server, and content-tool development of the digital murder-mystery experience "戏精大侦探", using Cocos, mini games, mini programs, Pomelo, MongoDB, Redis, Vue, and Element UI.
• Contributed to iOS app builds and App Store releases.