Project Overview
I developed an autonomous, serverless real-time concierge microservice to orchestrate group logistics, schedules, and pedestrian navigation for our group attending Rock in Rio 2026, combining Google Cloud Run, Google AI Studio (Gemini), Google Calendar, and Maps directly into a WhatsApp group.
The Challenge
Massive music festivals present severe cell-tower congestion, multiple concurrent stages, and heavy friction in coordinating group schedules, food stops, and walking distances. The goal was to build a zero-marginal-cost assistant that handles dynamic schedule shifts and geolocation through conversational interfaces.
Technical Architecture
- Google Cloud Run (Serverless): Python/FastAPI container deployed in
southamerica-east1(São Paulo), providing 24/7 uptime, automated SSL termination, and instant webhook handling. - Google AI Studio (Gemini 3.5 Flash): Natural language reasoning engine classifying fuzzy inputs into discrete calendar actions (e.g., mapping conversational phrases like “still eating, let’s push it a bit” to an automatic 15-minute event reschedule).
- Google Calendar API: Bi-directional event synchronization across 15 festival shows. Webhook triggers allow participants to patch event times dynamically from the chat.
- Google Maps & Local Routing: Distance Matrix pedestrian routing coupled with an offline Haversine mathematical fallback to maintain functionality under network drops.
- Companion Mode Support: Built specifically to support linked Android devices (WhatsApp Companion Mode) using static GPS pins, single-tap hotspot buttons, and an HTML5 fallback web check-in (
/checkin). - Group Engagement: Scheduled hourly festival trivia quiz with live leaderboard scoring before headliner sets.
Key Takeaway
This project demonstrates how Google Cloud’s serverless primitives and Gemini’s free-tier API quotas allow developers to build and deploy robust, multi-service event orchestration tools with zero infrastructure cost and minimal operational overhead.