From Idea to Launch: How I Built an Instant Messaging App on a Weekend

This is a submission for the New Year, New You Portfolio Challenge Presented by Google AI

About Me

I'm Ryan Guo, a backend developer and AI enthusiast. My goal for 2025 is to bridge the gap between ideas and tangible products through Python and AI-driven development workflows. I believe that with the right tools, even a single developer can build robust, scalable solutions at lightning speed.

OneLink is the manifestation of this belief—a cross-device, login-free real-time messaging app that I built in just one week by collaborating with Antigravity AI.

Portfolio

OneLink is fully deployed on Google Cloud Platform. You can experience the real-time file sharing and messaging below:

Live Demo (Frontend)

live demo

Backend Service (Cloud Run)

backend docs

• GitHub Repository: OneLink Source Code

---

How I Built It

Tech Stack

• Backend: FastAPI (Python), WebSocket (Full-duplex communication).
• Frontend: React 18, TypeScript, Vite.
• Infrastructure: Google Cloud Run (Backend logic), Google App Engine (Frontend hosting).

Development with Google AI Tools

I utilized Antigravity AI as my primary thought partner throughout the lifecycle:

1. Rapid Scaffolding: AI helped generate the initial FastAPI WebSocket manager and React state architecture, turning a concept into a working prototype in hours.
2. Performance Troubleshooting: When Base64 encoding caused memory spikes and UI lag, AI suggested switching to Binary Chunked Transfer, which became the backbone of the app.
3. Deployment Automation: AI streamlined the GCP integration by generating optimized Dockerfiles and gcloud deployment scripts, ensuring the service stayed within the Google Cloud Free Tier.

Design Decisions

I chose a Zero-Database Architecture. All messages and file chunks are handled in-memory and broadcasted instantly. This ensures maximum privacy (no data is stored on disk) and minimal latency.

---

What I'm Most Proud Of

The technical highlight I'm most proud of is the Binary Chunked Transfer Mechanism.

Transitioning from standard Base64 encoding to raw binary chunks (64KB each) was a game-changer. This optimization resulted in:

• Efficiency: A 33% reduction in payload size and a 40-50% increase in transfer speed.
• Stability: The ability to handle files up to 100MB without timing out the WebSocket connection.
• UX Excellence: I implemented a visual feedback system with animated progress bars (blue-purple for uploads) that keeps the UI responsive even during heavy data bursts.

Building OneLink proved that by leveraging AI assistance and Google Cloud’s serverless infrastructure, I can focus entirely on solving user problems rather than managing servers.