susan-awori/UMOJA-PAMOJA-CHAMA

# Umoja Pamoja AI Arbitrator > An AI-powered arbitration assistant designed for chama groups to resolve disputes fairly using the chama’s own bylaws, contribution records, and M-Pesa statements. --- # Table of Contents 1. Problem Statement 2. Solution Overview 3. Key Features 4. Tech Stack 5. System Architecture 6. Agent Architecture 7. Data Pipelines 8. How the System Works 9. Example Use Cases 10. Project Structure 11. Local Setup 12. Deployment 13. API & Interaction 14. RAG Implementation 15. Multilingual Support 16. Data Handling Policy 17. AI Neutrality Policy 18. Challenges Solved 19. Future Improvements 20. Team Members 21. Screenshots & Demo 22. License 23. Acknowledgements --- # 1. Problem Statement Many chama groups face conflicts involving: - Late contributions - Missing payments - Welfare fund claims - Merry-go-round payouts - Financial inconsistencies - Penalty calculations These disputes are usually handled manually through: - WhatsApp chats - Phone calls - Spreadsheets - PDFs - M-Pesa statements This often leads to: - Emotional arguments - Lack of transparency - Bias accusations - Poor record keeping - Delayed resolutions Most chama leaders do not have a centralized system that can: - verify contributions, - check rules, - and make fair decisions quickly. --- # 2. Solution Overview Umoja Pamoja AI Arbitrator is an AI-powered dispute resolution assistant built for chama groups. The system: - Reads chama bylaws - Understands disputes in English, Kiswahili, and Sheng - Verifies M-Pesa transactions - Checks contribution records - Calculates fines - Provides grounded rulings using the chama’s own constitution The AI acts as a neutral digital mediator. --- # 3. Key Features ## Core Features - AI-powered arbitration - Multilingual support - Bylaw citation engine - Financial verification - Contribution tracking - Fine calculation - RAG-powered retrieval - Structured + unstructured search - Explainable AI decisions --- ## Supported Langua …