iniFeng
Missions

Selected systems with the problem, role, stack, and operating detail exposed.

A compact archive of work across agent workflows, Cloudflare-native CMS tooling, and performance-focused frontend systems.

The through-line is operational clarity: show system state, preserve maintainable code paths, and keep infrastructure complexity behind clear product boundaries.

Mission

2025–Present

XMOV Bot Gateway

AI-driven agent gateway for code review and planning, with GitLab MR automation, Feishu callbacks, and execution-state visualization.

Reduced manual review effort and created a scalable foundation for risk screening, review aggregation, and PRD-aware planning workflows.

Problem Engineering teams needed AI review and planning workflows that were observable, bounded, and connected to real GitLab/Feishu operations.

System highlight Serial, observable agent execution with honest provider capability display and review result persistence.

Active production systemTypeScriptReactNode.jsHonoGitLabFeishuAI Agents
Clear product framing. Controlled operational overhead.Archive

Mission

2026

Blog & CMS on Cloudflare

Edge-native blog and CMS system using Workers, D1, R2, and KV, with Markdown authoring, admin workflows, and hybrid static/dynamic rendering.

Built a lightweight publishing foundation with low-latency delivery and clear operational boundaries for a personal engineering console.

Problem The portfolio needed to become a durable publishing system without splitting into a separate backend or heavy CMS product.

System highlight Single TanStack Start app with server-side Cloudflare bindings, D1 posts/settings, R2 assets, and admin safety controls.

Live foundation, iteratingTypeScriptTanStack StartCloudflare WorkersD1R2KVCMS
Clear product framing. Controlled operational overhead.Archive

Mission

2021–2024

SSR Performance & Admin Systems

Built Vue and React H5/admin systems, handled complex forms and state logic, and optimized SSR render paths for production surfaces.

Improved frontend observability and first-load performance through render-pipeline optimization, blocking-request cleanup, and Sentry integration.

Problem Production H5 and admin surfaces needed faster first loads, clearer monitoring, and maintainable state-heavy workflows.

System highlight TTFB/LCP investigation, request-chain cleanup, SSR render path tuning, and monitoring-driven regression analysis.

Archived experience baseVueReactNuxtTypeScriptLighthouseSentrySSR
Clear product framing. Controlled operational overhead.Archive