Hi, I'm Dang Nguyen.
After hours, I turn small curiosities into games and tools. Then strangers actually download them. This is where they live.
The ones that made it out the door. Live on the App Store, Google Play, and the web, with real people using them.
Box Push: 1000 Levels PuzzlePush every crate onto its goaliOS / Android
Pixel Paint: Color by NumberTap the numbers, reveal the pictureiOS / Android
Dot Shooter: Space Bullet HellTwin-stick bullet hell rogueliteiOS / Android
Good Boy: Cozy Herding PuzzleYou are the dog; sheep flee where you stepiOS
Tidy Match: Triple Match 3DCozy animal tile-match puzzleiOS / Android
Orbit Sort: Gem Ring PuzzleSpin rings, sort the gemsiOS / Android
Ascent: Reverse Block PuzzleBlocks rise instead of falliOS / Android
Idiomix: Learn VietnameseWord puzzle with real idiomsiOSReal code, half-built, waiting on the unglamorous last mile. Usually I get pulled toward the next idea before the current one ships.
Live speech translation on your phone, for Vietnamese, Japanese, and English.
Fully coded and passing tests. It's stuck at the boring last mile: real-device testing for accuracy and battery, plus the signing dance to ship. The hard part is done. The finishing isn't.
A document scanner: capture, crop, clean up, OCR the text, sign, export to PDF.
The whole core works: scan, multi-page, filters, copy text, signatures, PDF. What's missing is the unglamorous shell, monetization and store polish, and the focus to push it over the line.
Scan a real space with iPhone LiDAR and share a 3D walkthrough as a web link.
Built for property and resort tours: scan a room, get a browser walkthrough with a minimap, no 3D skills needed. Phase 1 runs clean in the simulator. It's waiting on real-device LiDAR testing, which is exactly where I keep not finding the time.
AI mock interviews by phone: a voice picks up, interviews you like a real person, then hands back honest feedback.
Works end to end: a real voice interview by phone, in character, that ends when it should and gives honest feedback after. It is shelved on economics, not code. The realtime voice that makes it feel like a person costs too much per call to run at a price people would pay, and I have not found a stack that holds that quality while dropping the operating cost low enough to be worth shipping.
Projects I built, finished, and shelved. The ideas worked. The reason for anyone to pay did not.
A link-in-bio store that scraped your Shopee and TikTok Shop products into one page.
The MVP shipped. The buyers didn't. Serious affiliates already have their own tools, and the people doing it for fun, students and office folks, had no reason to pay. Even if they did, there was no sane way to sync hundreds of products at once, and nobody updates prices by hand. I built for a middle that turned out not to exist.
An AI agent to plan your whole Vietnamese wedding, stress-free.
It started as a plain seating planner. Then I got ambitious and rebuilt it into a full AI wedding planner, straight over the old code, which I lost in the process. Turns out couples don't want a robot running their big day, and a checklist in a spreadsheet does the job for free. The AI version died. The simple seating tool I had paved over is the part that came back.
A drag-and-drop seating planner for weddings: build the venue, drop guests onto tables, redo it every time someone cancels.
This is the original, the plain seating tool I built before I got ambitious and rebuilt the idea into WedPlan. WedPlan overreached and died; this one just works. I don't maintain it anymore, but it's free, runs on Firebase, and costs me nothing to keep up, so it stays online for anyone who still needs to seat a wedding.
A finder for VinFast EV charging stations.
It leaned entirely on VinFast's public API, which never exposed the one thing that mattered: which chargers are free right now. A station finder that can't tell you where you can actually plug in is just a map with pins. Without real-time availability there was no reason to pick it over anything else, and no way for me to add that on my own.
An AI chat widget a small business drops onto its site with one line of code, answering visitors around the clock and turning after-hours questions into leads.
It got users. It never got paying ones. The whole pitch was a bot that answers your visitors 24/7, but every platform now ships its own AI chat, and a small shop can paste a free one from its site builder or just point people at ChatGPT. Nobody had a reason to subscribe to a standalone widget, and with no paying users there was no reputation to trade on either. The product worked fine. The reason to pay for it stopped existing.
