Bengali + English · Retrieval-augmented generation
Multilingual RAG System
A multilingual retrieval-augmented Q&A system that answers Bengali and English queries from HSC Bangla 1st Paper textbooks with source citations.
PythonLangChainChromaDBFastAPIRedisDocker
The context
Bengali textbook question answering needs both meaningful text comprehension and a traceable connection to the source material.
The engineering
- Processed textbook PDFs using PDFMiner and OCR.
- Built a retrieval pipeline with LangChain, ChromaDB, and the OpenAI API.
- Served Bengali and English questions through FastAPI, with Redis and Docker in the stack.
The outcome
- Bengali and English query support
- Source-cited answers from textbook content
- A working application of low-resource NLP
Next projectPaediatric Orthopaedic Surgeon Practice Management System
↗