Rawhatur Rabbi

Rawhatur Rabbi

Research Assistant at Brac University, passionate about deep learning and computer vision.

AgriBuddy: AI-Powered Agricultural Guidance System

Overview

AgriBuddy is a multilingual, AI-powered support platform designed to provide personalized agricultural guidance to smallholder farmers in Bangladesh.
The project was developed during the BRAC AI Hackathon (AgriBRAICution), combining computer vision, retrieval-augmented generation (RAG), and a smart chatbot to help farmers with crop disease detection, weather insights, and cultivation tips.

Key Contributions

  • Disease Detection: Implemented a CNN-based rice disease detector trained on the Paddy Doctor dataset, enabling real-time classification of 10 rice conditions.
  • Smart Chatbot: Integrated an agentic RAG framework using Bangla embeddings to deliver context-aware, localized answers in Bengali.
  • PWA Development: Designed a Progressive Web App with offline support, combining weather forecasts, soil analysis, and personalized cultivation guidance.
  • Team Collaboration: Worked with a multidisciplinary team spanning BRAC University, BUET, and DIU; co-authored a technical report and presented results publicly.

Tech & Tools

  • Deep Learning (CNNs) for crop disease detection
  • LangChain + RAG for chatbot integration
  • Bangla Embeddings for language support
  • PWA stack: React · Node.js · IndexedDB for offline mode
  • Cloud deployment for real-time data access and scalability

Impact

  • Brings accessible AI guidance to smallholder farmers in Bangladesh.
  • Improves early disease detection, reducing crop losses.
  • Provides a localized, farmer-friendly interface in Bangla for maximum adoption.
  • Sets the stage for scalable agricultural advisory platforms in other regional languages.