emanikor/ConnectTurkana

Turkana agency Tech 🚀 Mifugo: AI-Powered Livestock Market Intelligence Mifugo is a localized, AI-driven economic dashboard designed to eliminate pricing asymmetry in Turkana County's open livestock markets. By integrating local Large Language Models (LLMs) with a robust microservice architecture, Mifugo allows NGOs and local officials to capture real-time pricing data, generate autonomous market analyses, and protect pastoralists from market exploitation—all designed to operate entirely offline in low-connectivity environments. ## 🏗️ Architecture & Tech Stack This project is built using a modern, containerized AI-engineering stack: * **Frontend:** React.js (Clean, user-centric dashboard for market visualization) * **Backend:** Nodejs/python / FastAPI (High-performance, asynchronous API bridge) * **Database:** PostgreSQL (Relational data storage, containerized) * **AI Engine:** Ollama / Llama 3.2:1b (Local LLM inference for offline market analysis) * **Infrastructure:** Docker & Docker Compose (Fully containerized microservices) ## ✨ Core Features * **Local AI Inference:** Utilizes Llama 3.2 running locally via Ollama to analyze raw Postgres data and generate 2-sentence actionable business advice without requiring expensive cloud API calls. * **Data Ingestion API:** Secure REST endpoints built with FastAPI and Pydantic validation to receive and sanitize livestock data from field enumerators. * **Containerized Microservices:** The entire database, API server, and AI engine are isolated in Docker containers, ensuring the system can be deployed seamlessly to any environment. * **User-Centric UI:** A dark-themed, high-contrast dashboard designed for quick data scanning and AI insight delivery. ## 📸 System Previews * **Dashboard View:** ` ` * **AI Terminal Output:** ` ` ## 🚀 How to Run Locally Because the entire architecture is containerized, deployment is completely reproducible. 1. **Clone the repository:** ```bash git clone github.com cd mif …