Yo, listen up—here’s a story about how I’ve been building apps this year, one phase at a time. I’m BLUE Copper (Yuri Saito).
Everything here started in this fiscal year, and each project became a stepping stone to the next.
I want to write this down not just as a changelog, but as a story of how my tech stack grew and evolved.
Phase 1: React Native Expo
Garakta
This was my very first attempt with React Native Expo. The concept was messy, unclear, and honestly just me trying to make something fun without thinking about App Store guidelines. Unsurprisingly, it kept getting rejected.
But still—thanks to Garakta, I got comfortable with React Native itself. That learning was incredibly valuable later on.
Wander Alley
Basically Garakta, but with a new name and a clearer concept. I added a few extra features, refined the idea, and finally managed to get it through App Store review.
Here, I realized how important communication and product design really are—not just coding. Deciding what to include and what to leave out was a big marketing/consulting-type lesson.
Switchery
My first app where I deeply cared about concept. “Switch your mood like a magical transformation item.”
I invited my mom to test via TestFlight and shaped the app with her feedback. It even includes original music I composed for it, which makes me really happy.
Array Mbira Guide
This one came from my mom, too—she had bought an Array Mbira about a year earlier and wanted to practice. I suddenly thought: “Wait, I can build an app for that now!”
I used her actual recordings, cut them up automatically with a Python script, and made a simple guide app. Honestly, the behind-the-scenes code to process the sounds was more work than the app itself. But I did it in about a day.
Phase 2: Socket Communication and Q-Learning
Open Trade Poker
This card game was actually something I invented at home three years ago. Back then, I tried Socket.io but failed—it was too difficult. I ended up making a simple single-screen web app instead.
This year, I revisited the game. The rules were already designed, so I just needed to implement them. It only took me one day.
But then I realized: matchmaking is useless without enough players. I needed AI opponents.
First, I tried a rule-based “optimal strategy” and trained against that. My mom tested it—and crushed the AI every time. My heart broke a little.
Then I studied her strategy, retrained the model with it as a teacher, and finally got an AI that could stand strong.
(If you’ve got even stronger strategies, please reach out—I’d love to improve it further.)
Phase 3: “Wait, I Can Draw with Code?” and My VTuber Dream
SVGTuber
A couple of months before this, I had recreated my drawing style using Python code. That inspired me to port it into a web app.
With React × TypeScript × SVG × MediaPipe, I made a resolution-independent VTuber without Live2D—just code.
Why? Because I’m terrible at precise hand drawing and fine-tuning Live2D rigs. I’m lazy. Code-based parameter tweaks were good enough for me.
Someday, I want to stream in this form.
Phase 4: Overcoming My Database Fear & First Steps with Vercel
Pixiome
This was inspired by Swarm Sketch and r/place. I love Minecraft, so I designed a multiplayer system where players shape a biome together—raising land, adding water, planting trees.
It was my very first time handling databases and APIs. I was nervous. Honestly, I still don’t fully understand some of it. My daily batch job is broken right now—I’ll need to fix that soon.
But deploying on Vercel was surprisingly easy: connect Git, set environment variables, done.
Phase 5: OpenAI API Integration
Hashtag Suggester
This one came from a personal pain point: I’m terrible at thinking of hashtags for my posts. I kept asking AI for suggestions anyway, so I thought—why not make a web app for it?
Thanks to my prior experience with Vercel on Pixiome, integrating OpenAI API was super smooth. Just drop in the API key as an env variable. Easy.
Phase 6: First Flutter App
Influencer Goal Maker
Funny thing—I had already made a follower-growth simulator two years ago in React Native. It calculated monthly goals using a logistic curve.
This year, since I needed Flutter experience for my main job, I decided to rebuild it entirely in Flutter. And since I had just built Hashtag Suggester, I added that feature in too.
The backend? Same shared Vercel setup.
And now—it’s on the App Store! 🎉
What’s Next
Right now, I’m working on Bead Designer—a tool for creating beadwork layouts with rings and center beads.
Every phase taught me something new, from Expo to Socket.io, from databases to Flutter.
And honestly, the most exciting part is that I can now share these projects with the world.