Shwe Taungthuu
Gemini 2.5 Flash-powered bilingual agricultural advisory agent for Myanmar farmers — Google Build with AI MMDT 2026
| Google Developers Community Build with AI MMDT 2026 | 2026 |
AI-powered bilingual agricultural advisory app for Myanmar farmers
Overview
Built a full-stack agricultural advisory application enabling Myanmar farmers to ask crop and pest management questions via Burmese voice or text — with structured advisory responses generated in real time using Google Gemini 2.5 Flash via Vertex AI.
Backend
- Python + FastAPI backend deployed serverlessly on Google Cloud Run
- Used Application Default Credentials (ADC) for keyless, production-grade cloud authentication — eliminating hardcoded API keys and enabling secure service-to-service calls without credential exposure
Frontend
- Built with Tailwind CSS, Lucide Icons, and Web Speech API for browser-native speech-to-text
- Implemented a Voice History feature allowing farmers to replay original voice queries inline within the chat
- Designed for low-literacy, field-usability contexts
Development Workflow
- Developed the entire application using Gemini CLI in Google Cloud Shell — AI-assisted coding workflow from architecture to deployment within the Google ecosystem
Tech Stack: Python · FastAPI · Google Gemini 2.5 Flash · Vertex AI · ADC · Google Cloud Run · Web Speech API · Tailwind CSS · Lucide Icons · Gemini CLI