Year — 2026
Agri-assistant-RAG
Agri Assistant is a production-style agricultural advisory platform built for location-aware crop guidance and document-grounded farm decision support. The project combines a FastAPI backend, retrieval-augmented generation, structured agronomic analytics, and a React frontend into a single end-to-end system for agricultural Q&A.
Stack
Python 3.12+FastAPIPydantic + Pydantic SettingsLangChain and LangChain CommunityGroq LLM integration via langchain-groqHuggingFace embeddingsFAISS vector searchPostgreSQL + pgvectorRedisPyPDF, pandas, PyYAML, httpx, requestsReact 19ViteTypeScriptAxiosLucide React
Key Features
- ✨Answers farmer questions using indexed PDF knowledge bases
- ✨Evaluates agronomic CSV datasets for seasonal and production insights
- ✨Checks climate and weather conditions for local recommendations
- ✨serves contextual, location-aware answers for Kenyan counties and farm areas
- ✨exposes health checks, caching, request tracking, and API endpoints for integration
- ✨Logs interactions and confidence signals for traceability.
Technical Highlights
- Retrieval-augmented answering from domain documents in data/pdfs
- Weather-driven guidance based on latitude and longitude using the Open-Meteo API
- CSV analysis for production and farm outcome data
- Per-request logging and request IDs for tracing
- Redis-based caching and short-term session memory
- Health monitoring for database, cache, and embedding services
- County-based UI selection for Kenyan regions
- Multi-audience response handling for farmers and experts
- English and Swahili output support in the API request schema
My Role
- Implemented the project fully