Inspiration
Building on my success from last year using Firebase for hackathon tools, I wanted to push the boundaries of "vibe coding" this year. I challenged myself to see if I could co-create a high-performance application entirely through natural language interaction using Google AI Studio and Gemini.
What it does
The Team USA Archetype Agent is an immersive AI-driven experience that allows fans to discover their "Olympic Mirror." By analyzing biometric traits and athletic abilities, the agent scans a custom dataset of 120 years of Team USA history to find historical peers across Olympic and Paralympic disciplines. It doesn't just find a match; it explains the why—connecting your physical profile to the mechanical demands of elite sport. It treats olympic and Paralympic at same level of analysis.
How I built it
This app was built using a Zero-Code / Agentic workflow over a single weekend: Engine: Powered by Gemini 3.1 Flash and Gemini 3.0 via Google AI Studio. Data: my BigQuery project containing historical Team USA athlete profiles, curated from teamusa.com. Deployment: Containerized and deployed to Google Cloud Run. Vibe Coding: I acted as the Architect and Product Owner, providing vision and feedback, while Gemini acted as the Lead Engineer to implement the full-stack logic.
Challenges I ran into
The biggest hurdle was data signal vs. noise. While I successfully scraped legacy data, the historical records were often sparse or inconsistently formatted. I had to refine the AI's "Intelligence Layer" to handle these gaps gracefully, ensuring that even with limited data, the user receives a technically accurate and inspiring analysis.
Accomplishments that I'm proud of
I am incredibly proud to have completed a fully functional, full-stack application without writing a single line of code manually. By iterating through feedback loops with Gemini, I was able to build a tool that captures complex athletic archetypes and historical context in a way that feels intentional and polished.
What I learned
AI Studio is a force multiplier. I learned that the future of development isn't just about syntax; it's about architectural clarity. To build a successful AI app, you must understand how to integrate data backends (GCP/BigQuery) with frontend design, and how to guide an LLM through the "last mile" of deployment and permissions management.
What's next for Team-USA-Archetype-Agent
The roadmap includes integrating Metabolic Intelligence. I plan to source higher-fidelity physiological data to match fans based on metabolic rates and energy systems, making the "archetype" matching even more scientifically robust.
Built With
- cloudrun
- geminiapi
- google-bigquery
- googleaistudio
Log in or sign up for Devpost to join the conversation.