Full-Stack Engineer · Tokyo · Remote (GMT+9)

Kai Li

Full-Stack Engineer  ·  ~8 yrs  ·  Web → Flutter  ·  AI-native

Open to remote roles  ·  currently at ELESTYLE, Tokyo
React / TypeScript Flutter AI Agents · Tool Calling Node.js / Nest Real-time / High-concurrency Fintech / Payments Open-source contributor Highly-Skilled visa
01

About

Full-stack engineer focused on AI-agent systems and real-time, high-concurrency products — the exact surface an agentic trading product lives on. At ELESTYLE (Tokyo fintech) I build a PCI-DSS v4.0 payment platform serving 200k+ merchants nationwide, going deep on MQTT real-time messaging and PostgreSQL / Redis data modeling with reliable delivery under load.

I ship AI agents end to end. My flight-monitor agent uses GPT-4o as its decision core (Tool Calling + Function Calling) to watch multi-source pricing via Playwright, store history in TiDB, and push alerts through a Telegram bot — Dockerized, with an MCP server. It's the same monitor → analyze → recommend → alert loop a trading agent runs.

Web to mobile — exactly this role: React / Next.js on the web and Flutter on mobile. I also design, build, and operate products solo — StarFlix (star-flix.cc) is a Next.js full-stack streaming platform I run end to end, from product to ops.

AI-native by default: I work daily in Claude Code / OpenAI Codex / VS Code, with AI tooling fully woven into my workflow — so "proficient with AI coding tools" isn't a claim, it's already how I ship.

Active open-source contributor — merged PRs to yt-dlp (90k+ ★), vitejs/vite-plugin-react, and cloudflare/vinext; author of unplugin-version-injector on npm. Fluent technical English; overlaps well with GMT+8 hours from Tokyo.

Experience~8 yrs
Leadership5 yrs
Based inTokyo, Japan
Time zoneGMT+9 · remote-ready
GitHub repos160+
LanguagesEN · 中文 · 日本語
VisaHighly-Skilled Pro
02

Experience

ELESTYLE Inc. Dec 2024 — Present
Full-Stack Engineer
Leading Japanese fintech · elepay powers payments for 200k+ merchants nationwide · clients incl. SMBC, Kirin, Coca-Cola · PCI-DSS v4.0 / ISO 27001
  • Full-stack development of core SaaS products — oneQR / Wait / KDS / Dashboard
  • Solved an MQTT reconnect request-storm in KDS with throttling + a Redis-persisted dedup window, guaranteeing message consistency and reliable delivery under high concurrency
  • Refactored the Wait module — +50% runtime performance, +11% container resource efficiency
  • Designed and tuned PostgreSQL schemas for payment ledgers and merchant config — 40%+ faster responses
  • Core dev on the SMBC × elepay scheduled-billing project: owned the frontend, went deep into the Java backend, drove the test loop and part of the business-rule design — shipping a bank-grade compliant payment flow
  • Delivered Webhook event-type refactor, Merchant Keys API integration, and payment-URL expiry
  • Used behavior analytics to proactively detect (not user-reported) and fix a production session-loss bug — data-driven debugging
ReactVue TypeScriptJava MQTT PostgreSQLRedis PCI-DSSSMBC
Hangzhou EasyDataInsight (DTStack) Jan 2021 — Dec 2024
Senior Frontend Engineer & Frontend Lead
Big-data SaaS · EasyV digital-twin visualization platform · grew into the group's largest, most stable cash-cow product
  • Built the platform 0 → 1 — owned architecture and implementation of the editor, blueprint, and asset modules
  • Self-initiated and built an AI support agent 0 → 1 (RAG knowledge base + multi-turn dialogue + Tool Calling): zero operating cost, +20% CSAT, −30% headcount
  • Multi-layer performance work (HTTP/2, CDN, Service Worker, lazy-loading, data pre-cache) — +60% first paint
  • Designed fine-grained button-level RBAC with multi-tenant isolation
  • Decoupled the codebase into a monorepo of feature sub-modules, sharply raising dev velocity
  • Managed the frontend team — hiring, mentoring, and on-site delivery when projects stalled
ReactTypeScript WebGLNest.js RAGAgent Team Lead
Zhejiang Yingshu Biotech May 2018 — Dec 2020
Frontend Engineer → Frontend Lead
  • Rebuilt the H5 stack with SSR + skeleton screens for a markedly faster first screen
  • Integrated all payment rails (WeChat / Alipay / LianLian) across both B- and C-side flows
  • Standardized engineering — ESLint + code review; migrated the mini-program from mpvue to Taro
  • Built a Node.js + Puppeteer automation crawler: scheduled scrape → Excel → auto-email
  • Owned frontend tech selection and base architecture across PC / H5 / mini-program / POS
VueReact TaroNode.js Mini-program
03

Skills

Frontend
ReactNext.js TypeScriptWebGL Taro
Cross-platform
FlutterElectron TauriMini-program
AI Agents & Automation
GPT-4oClaude API Tool CallingFunction Calling RAGPlaywright Telegram Bot Claude CodeCodex
Backend & Services
Node.jsNest.js PythonFastAPI PostgreSQLRedis MongoDBDocker REST API
Build Tools
ViteWebpack Rollupunplugin monorepo
Engineering Leadership
Tech selectionTeam building StandardsCI/CD
04

Open Source & Personal Projects

05

Contact