Photo of Nguyen Tuan Thanh Ho Chi Minh City8+ years with ReactFrontend Lead

Nguyễn
Tuấn Thanh

I lead frontend teams and build real-time AI interfaces with React and Next.js.

Try the console. It streams like the AI products I build.
Ask my portfolioReady
Selected work

What I build, running live

Small working versions of what I ship at Pillarlab, the AI unit of Ringkas. Click around.

Embeddable chat SDK

An iframe embed of about 22 KiB, so partners such as Honda run our AI assistant on their own websites, with its storage isolated from theirs.

iframe SDKSeparate bundleSecurity
partner-website.example
Partner
Click the bubble

Multi-tenant frontend

One build serves every partner program. Branding, theme and locale resolve per entry at runtime.

Runtime themingi18nRTL
Aurora Bank
Hi, how can I help you today?Ask about home loans, documents or your application.Start
resolveTenant(entry) → {
  tenant: "aurora",
  locale: "en", dir: "ltr",
  theme: { primary: "#4f6bff" }
}

Streaming that reads smoothly

AI replies arrive over SSE in bursts. A typewriter buffer turns them into steady text.

SSEAG-UIrAF
Raw SSE chunks

With the typewriter buffer

0 bursts received

A team where every AI agent follows the same rules

A CLAUDE.md / AGENTS.md rule book shared by Claude, Codex and Cursor, 16 living architecture docs, and lint that enforces the boundaries. Teammates update the rules in the same merge request as their code.

Feature-Sliced DesignESLint180+ Vitest suites
# How we build here
Shared by Claude, Codex and Cursor (AGENTS.md).

- Layers: app → pages → widgets → features → entities → shared
- Import downward only. ESLint fails the build otherwise.
- Read docs/architecture/<topic>.md before touching <topic>.
- Changed a convention? Update this file in the same MR.

Illustration of the setup, not the company's code.

Experience

8+ years of shipping React products

Frontend lead with 8+ years building React and Next.js products, including 5+ years leading a frontend team of 12+ engineers. I combine hands-on architecture (real-time AI chat over SSE, multi-tenant platforms, embeddable SDKs) with growing engineers, setting standards and turning business goals into delivery plans. Lately I build the operating model for AI-assisted teams: machine-readable rules, living architecture docs and lint-enforced boundaries, so every engineer's AI agent ships code that follows the same conventions.

  1. Pillarlab AI, Ringkas's AI unitSince March 2026

    • Started the Pillarlab frontend codebase from scratch and set up the team's AI-assisted development model: a CLAUDE.md / AGENTS.md rule book shared by Claude, Codex and Cursor, plus 16 living architecture docs the agents load on demand.
    • Backed the written rules with tooling: Feature-Sliced Design import boundaries in ESLint, a cycle check in lint and 180+ Vitest suites. Teammates update the rules in the same merge request as their feature code.
    • Designed and built the AI chat client on AG-UI over SSE: streamed text, tool calls and generative UI cards, with a typewriter buffer that smooths the backend's bursty token stream.
    • Built the multi-tenant frontend: one build serves every partner program, with branding, theme and locale resolved per entry at runtime.
    • Built the Program Chat SDK iframe embed (about 22 KiB) so partners such as Honda run the assistant on their own websites.
    • Kept app.pillarlab.ai fast as conversations grow: lazy-loaded routes, a separate embed bundle, scroll-up history pagination and server-state caching in React Query.
    • Owned frontend security: XSS and data-exfiltration hardening of model output, in-memory tokens, isolated embed storage.

    Mortgage platform for Indonesia and Saudi Arabia

    • Delivered mortgage features for Saudi Arabia with partners ROSHN and SNB, gradually splitting country-specific code out of the legacy shared base so each market ships independently.
    • Built document, KYC and consent flows on AWS S3, and defined API contracts with the backend teams.
    • Also shipped backend work on Node.js and GraphQL services (AWS Lambda, PostgreSQL) with unit tests.
    ReactNext.jsTypeScriptViteZustandReact QueryTailwind CSSSSEGraphQLNode.jsPostgreSQLPythonAWSServerlessSentry
  2. Led a frontend team of 12+ across six client products: task planning, onboarding and day-to-day technical guidance.

  3. Built core UI features with React and Redux, integrating REST APIs from the Java Spring backend team.

App Store

Apps I make and use every day

Small, focused Mac apps. Each one started as a problem of my own.

Toolbox

What I work with

Leadership

Leading a 12+ engineer teammentoringcode-review standardstechnical directionestimation and delivery planningstakeholder communication

AI engineering

Claude CodeCodexCursorCLAUDE.md / AGENTS.md rule booksagent skillsAI code reviewlint-enforced conventions

Frontend

ReactNext.jsRemixViteReact NativeAngularChrome extensions

State and data

React QueryZustandRedux ToolkitApollo ClientSSE streaming

Architecture

Feature-Sliced Designmulti-tenant themingembeddable iframe SDKsAPI contract design

Performance

Code splittinglazy loadinglist virtualizationrender optimizationbundle-size budgetscaching

Quality and security

VitestTesting LibrarySentryGTMMixpanelXSS hardeningsecure token handling

UI

Tailwind CSSshadcn/uiMUIAnt DesignStyled Components

Backend and cloud

Node.jsNestJSGraphQLPostgreSQLRedisSQSFastAPIAWSServerless

Languages

TypeScriptJavaScriptPythonSQL
Photo of Nguyen Tuan Thanh

Looking for a frontend lead?

I'm based in Ho Chi Minh City, Vietnam and fluent in English. Email is the fastest way to reach me.