YieldSync
AI-powered agricultural consultant supporting multiple native languages.
Architecture & Overview
Farmers often lack immediate access to expert agronomy advice, real-time market data, or accurate weather forecasts in their native languages. To help solve this, I engineered a highly responsive AI agent capable of communicating fluently in Sinhala, Tamil, and English. The system features real-time crop disease diagnosis, live weather forecasting, and market price tracking via custom tool-calling powered by the Llama-3.3-70b-versatile model. The backend utilizes Python and FastAPI via the AgentKernel framework, while the frontend is built with Next.js. This required navigating complex CORS security policies and mastering function-calling so the LLM could execute native Python code for real-time data fetching.