Tran Tuan Hiep — Full-Stack & AI Engineer

A Mit-Suu PicturePresents

a full-stack & AI engineer story

Starring: React · Next.js · Node.js · TypeScript
Filmed on location in: Da Nang, Vietnam
Graduating: December 2026

EXT. DA NANG — GOLDEN HOUR

The Protagonist

Portrait of Tran Tuan Hiep — Casting still · Take 3Tran Tuan Hiep

I build web products end to end — from the first vague requirement to production on Azure. React and Next.js on the front, Express and MongoDB behind it, tests holding it all together, and lately, AI that proposes while the system decides.

Based in: Da Nang, VN
Studying: Software Engineering · FPT University
Graduating: Dec 2026
Currently: AI Engineer Intern @ Tekmium

MONTAGE — FROM BANKING SYSTEMS TO AI AGENTS

The Journey

ActI

INT. FPT JAPAN SOFTWARE — BANKING FLOOR — DAY

Software Engineering Intern

FPT Japan SoftwareSep 2025 — Jan 2026Vietnam · On-site

IBS, a Japanese banking system. A team of ~20 inside a ~500-engineer project spread across Vietnam, Japan and the Philippines.

Shot 01Analyzed Java Spring Boot source code and enterprise system configurations.
Shot 02Wrote technical documentation based on source code analysis.
Shot 03Used SVN for source retrieval and Git for version control; Jira, Confluence and Excel in an Agile process.
Shot 04Collaborated with developers, QA, BA and BrSE in a distributed international team.

Java
Spring Boot
SVN
Git
Jira
Confluence

ActII

INT. TEKMIUM — AI LAB — NIGHT

AI Engineer Intern

TekmiumAug 2026 — PresentDa Nang · Hybrid

Built SignalCoach, a full-stack interview-practice product with an AI interviewer that streams its answers in real time.

Shot 01Next.js 16 front end on Vercel and an Express 5 + TypeScript + MongoDB API on Azure App Service.
Shot 02Layered API (routes, controllers, services, models) with Zod validation, JWT in httpOnly cookies and one consistent error format.
Shot 03Per-turn agent loop — judge, pure state machine, persona — streamed over SSE, with a self-checking scorecard that cites the conversation.
Shot 04Single LLM gateway with multiple API keys, model fallback on overload and retries.
Shot 05Integration tests with supertest and in-memory MongoDB; a mock LLM provider keeps them deterministic.

Next.js 16
Express 5
TypeScript
MongoDB
SSE
Azure

Scene 03 · Now showing

Feature

Presentation

3 productions · 2026 · Team Leader on every one

Feature 01 / 03Capstone · AI · Team of 5

FlintFlow

AI Business Analyst System

Guides users from vague requirements to a full SRS with diagrams and Word/PDF export. The AI proposes; the system decides.

809: backend tests
211: frontend tests
85.5%: RAG correctness

Team LeaderJun 2026 — Present

Live ↗Source ↗

Feature 02 / 03Dev Tool · Research · Team of 2

APILens

REST API Smell Detection

A rule engine that reads your Express routes, OpenAPI docs and Postman collections — and tells you where your API smells.

9: smell rules
157: endpoints benchmarked
90%: human agreement

Team LeaderJun 2026 — Aug 2026

Live ↗Source ↗

Feature 03 / 03E-commerce · Team of 4

KIX Sneaker

Sneaker E-commerce Platform

A sneaker store from catalogue to checkout — and a Top 10 finish out of 120 teams in a UX/UI competition final.

Top 10: of 120 teams
4: core modules
4: person crew

Team LeaderApr 2026 — Jul 2026

Live ↗

Coming soon

More reels in the vault

github.com/mit-suu ↗

INSERT — THE NUMBERS BEHIND THE REELS

Box Office

1,020

automated tests

809 backend + 211 frontend on FlintFlow

85.5%

answer correctness

RAG knowledge chat — vs 16.4% without RAG

96.9%

Recall@3

Hybrid retrieval, best configuration

products led

Team Leader from requirements to production

FlintFlow knowledge chat · answer correctness on the evaluation set

With RAG85.5%

Without RAG16.4%

INT. THE STACK — EVERYONE ON SET

The Cast

React✦Next.js✦TypeScript✦Node.js✦Express✦MongoDB✦React✦Next.js✦TypeScript✦Node.js✦Express✦MongoDB✦React✦Next.js✦TypeScript✦Node.js✦Express✦MongoDB✦React✦Next.js✦TypeScript✦Node.js✦Express✦MongoDB✦

Redis✦BullMQ✦Docker✦Azure✦RAG✦Playwright✦Redis✦BullMQ✦Docker✦Azure✦RAG✦Playwright✦Redis✦BullMQ✦Docker✦Azure✦RAG✦Playwright✦Redis✦BullMQ✦Docker✦Azure✦RAG✦Playwright✦

Front-end

React
Next.js
TypeScript
JavaScript
Tailwind CSS
next-intl
HTML
CSS

Back-end

Node.js
Express 5
REST API design
Zod
JWT & OAuth
BullMQ
SSE
Webhooks
Spring Boot

Databases

MongoDB
Mongoose
Atlas Vector Search
Redis
MySQL
SQL Server

Testing & Quality

Vitest
Jest
Testing Library
MSW
Supertest
Playwright

CI/CD & Cloud

Docker
GitHub Actions
Azure App Service
Azure Container Registry
Vercel
Git
SVN

AI & Tooling

LLM features
AI agents
RAG
Vercel AI SDK
Gemini
Claude Code
Figma

EXT. RED CARPET — FLASHBULBS

Laurels

Official selection2026

Top 10 of 120 teams

Web Design UX/UI Competition · Final Round

With KIX Sneaker, an e-commerce platform built by a team of four.

Official selection2026

90% agreement with humans

University Research · APILens

Smell engine evaluated on 157 hand-labelled endpoints from 14 open-source repos, 0.82 macro precision.

Production notes

FPT University

Engineer's Degree in Software Engineering

Premiere: Expected Dec 2026
GPA: 7.9 / 10
Capstone: FlintFlow
Languages: Vietnamese · English

A portfolio by

Tran Tuan Hiep

Written & Directed by: Tran Tuan Hiep

Produced at: FPT University · Da Nang

Starring: React; Next.js; Node.js; TypeScript

Supporting Cast: Express 5; MongoDB Atlas; Redis; BullMQ; Zod

Stunt Doubles: Vitest; Jest; Playwright; Supertest

Visual Effects: Tailwind CSS; GSAP; Lenis

Locations: Tekmium AI Lab; FPT Japan Software; Azure · Vercel

Special Thanks: Team FlintFlow; Team APILens; Team KIX; Claude Code

No servers were harmed in the making of this portfolio.
All tests passed. Most of them on the first take.

…or is it?

Now casting for the sequel

Your team could be Act III.

Available full-time · Da Nang or remote. Open to front-end, back-end, full-stack and AI roles.

trantuanhiep28122003@gmail.com

Tran TuanHiep

Software Engineering Intern