AI-native · 8 yrs · shipping daily
Saigon, Vietnam

Ship living systems end to end.

Full-stack product engineer. I ship apps, APIs, data, auth, payments, and AI features users actually touch, including SpeakEn (AI English Coach). Every layer, from the pixels to the model loop.

Claude CodeMCPOpenCodeIn-app AI
View workSpeakEn

8 years in production. Distributed systems at scale for millions of users. Now I ship my own products, use coding agents daily, and own every layer.

Hien Phan, full-stack engineer
Next.jsNext.js
TypeScript
React
Tailwind
SQLite
DrizzleDrizzle
Node.js
Auth.js
Stripe
OpenAI
GitHub
Scroll

Tools I build with

TypeScript
Next.jsNext.js
React
Tailwind CSS
shadcn/ui
Motion
Zod
Node.js
PostgreSQL
SQLite
DrizzleDrizzle
Cloudinary
Vercel
Resend
Auth.js
Stripe
OpenAI
Canva
GitHub
file_type_nginxnginx
Linux
For hiring managers

What you get when we work together.

A full-stack engineer who ships products, ships AI features users touch, and uses coding agents to move faster without giving up ownership.

Open to full-stack & product engineer roles

Full-stack Engineer · Product Engineer

Start a conversation
0

Years shipping

Production software

0

Live products

Public, under my name

0

Stack layers

UI → API → data → deploy

0

Vertical owner

No handoff theater

Live product

SpeakEn (AI English Coach) is a real AI speaking coach, not a mockup or pitch deck.

Faster shipping with agents

I use Claude Code, MCP, and OpenCode daily so I can own more of the stack myself.

Full vertical ownership

UI, API, SQLite, auth, payments, deploy. One engineer who closes the loop.

Scale, then indie

Distributed systems at scale, then solo products under my own name.

Featured product

A live AI product I designed, built, and ship.

SpeakEn (AI English Coach) is the proof: speech pipeline, coaching loop, sessions, and progress. Full stack, not a slide.

SpeakEn
SpeakEn, screenshot 1SpeakEn, screenshot 2SpeakEn, screenshot 3SpeakEn, screenshot 4SpeakEn, screenshot 5SpeakEn, screenshot 6SpeakEn, screenshot 7SpeakEn, screenshot 8SpeakEn, screenshot 9SpeakEn, screenshot 10SpeakEn, screenshot 11SpeakEn, screenshot 12
View larger
1 of 12 · product surface
Solo full-stack

SpeakEn

My current project. AI-powered speaking practice with real-time feedback, session tracking, and measurable progress.

  • Speech to transcript to LLM coaching in one session loop
  • Personas, scoring, streaks, and review for real practice
  • Deepgram + OpenAI wired into production UX
Next.jsTypeScriptReactTailwind CSSshadcn/uiMotionZodLucide
UIAPIDBAuthSpeech pipelineUX flows
Case studyAll work
AI work

I ship AI products. I also use coding agents every day.

Claude Code, MCP, and OpenCode for building. In products, AI is a feature users touch, like the coaching loop on SpeakEn.

How I build with agents

Claude Code

Coding agents

Multi-file features, refactors, and reviews with a coding agent in the terminal.

MCP servers

Tool-using agents

Agents call real tools: repos, browsers, docs, and APIs through Model Context Protocol.

OpenCode

Open agent runtime

Open agent workflows for building products without locking into one vendor.

In-app AI

Product features

Speech, feedback, summarization, and generation inside flows users open every day.

A real agent session

✔ Mapped speech → transcript → LLM coach path
✔ Wired Deepgram + OpenAI into product UI
✔ Added scoring, personas, and progress state
ℹ Captured live product surface for review

Agents help me design, implement, and review. I still own architecture, product decisions, and what ships to production.

AI inside products

Users feel the model. I own the product loop.

SpeakEn

AI speaking coach, end to end

  • Speech to transcript to LLM coaching in one session loop
  • Personas, scoring, progress, and flashcard review
  • Deepgram + OpenAI wired into a real product UI
Open product
Indie products on this hub

AI as a product feature, not a demo

  • Summarization, generation, and coaching inside real apps
  • Auth, data, billing, and deploy in the same surface
  • Shipped under my name so the work is public
See work
Studio map

Projects, systems, and how they connect.

Every project is a real product surface. Each node links to a live tool or a case study. System nodes show the shared infrastructure: auth, database, payments, deploy.

  • Click any project node to open its detail drawer
  • System nodes reveal the shared stack
  • Keyboard: Tab to navigate, Enter to select, Esc to close

Studio Map

● apps○ coming○ systemsclick to explore
Experience

Building software for 0+ years.

From distributed systems at scale to freelance full-stack to shipping my own products.

2019-2022Large-scale messaging platform

Software Engineer

Built and maintained backend services using Java and ThriftRPC. Worked on messaging infrastructure at scale. Millions of users, real-time communication, distributed systems.

JavaJavaScriptThriftRPCDistributed Systems
2022-2027Freelance

Full-stack Developer

Delivered end-to-end web products for clients across industries. Full-stack ownership from UI design to database schema to production deployment.

Next.jsPostgreSQLTypeScriptReactTailwindNode.js
2027-PresentIndie Maker

Solo Founder

Shipping products under my own name. Building SpeakEn (AI English Coach) and other full-stack SaaS products. From idea to deployment, end-to-end ownership.

Next.jsSQLiteAuth.jsStripeAI/ML
Scrapheap

Things that didn't work. And what I learned.

I've started and failed more side projects than I can count. Each one taught me something I later turned into a post. Honest postmortems, not polished case studies.

Projects

Products I have built and shipped.

Real product surfaces with UI, data, deploy, and often AI features. Not mockups.

0

shipped

0

in lab

Shipped · Archived

V

VidSummify

Archived

#

Paste any YouTube link, get a concise AI summary emailed to you. Built it for myself but never found product-market fit.

Month 1
C

ClipGrid

Archived

#

Drag-and-drop calendar for scheduling social posts. Got a few users but churned. Too much competition from established tools.

Month 2
R

ReadKit

Archived

#

Save articles, get auto-tagged summaries, weekly digest email. Validation failed. Users wanted Pinterest, not a reading tool.

Month 3

In the pipeline

App #5

Month 5

Another product surface in the pipeline. Stay tuned.

App #6

Month 6

Halfway to 12. Each app is a learning milestone.

App #7

Month 7

More products, more ownership, more craft.

App #8

Month 8

Each month, a new product surface goes live.

App #9

Month 9

Nine months in. Learning what sticks, shipping fast.

App #10

Month 10

Double digits. Getting closer to the one that works.

App #11

Month 11

One more before the final. Learning every cycle.

App #12

Final milestone

The long-term vision: ship great software, end-to-end.

Open to full-stack & product engineer roles

Work with the engineer who ships the product and the AI inside it.

Full-stack ownership. AI product features. Daily use of Claude Code, MCP, and OpenCode. Clear communication. Ready for product-minded teams.

Full-stack EngineerProduct EngineerFounding Engineer (select)
harrisonphan5@gmail.comLinkedInSaigon, Vietnam · GMT+7 · remote-friendly