Project Report
Uploaded by
irsaf819Project Report
Uploaded by
irsaf819PROJECT REPORT
==============
Title: Multilingual Agentic RAG Chatbot System
Student: Irsaf Fatima
Date: June 2026
1. OBJECTIVE
------------
Design and develop a Generative AI powered chatbot using Retrieval Augmented
Generation (RAG) with multilingual support, dynamic document ingestion,
conversational memory, and agentic AI capabilities.
2. TECH STACK
-------------
- Frontend: Streamlit
- LLM: Groq (LLaMA 3.3 70B)
- Vector DB: ChromaDB
- Embeddings: HuggingFace (all-MiniLM-L6-v2)
- Framework: LangChain
- PDF Export: ReportLab
- Language Detection: langdetect
3. FEATURES
-----------
- Multilingual Support: English, Urdu, Arabic, Spanish
- Runtime PDF Upload and RAG
- Conversational Memory with context
- Agentic AI: Research Paper Generator
- PDF Export of generated papers
- Chat Sharing: WhatsApp, LinkedIn, Email
4. SYSTEM ARCHITECTURE
----------------------
User -> Streamlit UI -> LangChain RAG Chain -> ChromaDB -> Groq LLM -> Response
5. TEST CASES
-------------
| Test | Input | Expected Output | Status |
|----------------|--------------------|------------------------|--------|
| PDF Upload | Any PDF file | PDF ready message | Pass |
| English Query | "What is AI?" | English answer | Pass |
| Urdu Query | "AI kya hai?" | Urdu answer | Pass |
| Arabic Query | Arabic text | Arabic answer | Pass |
| Spanish Query | Spanish text | Spanish answer | Pass |
| Research Paper | "AI in Islam" | PDF downloaded | Pass |
| Chat Share | Click WhatsApp btn | Opens WhatsApp | Pass |
| Memory Test | Follow up question | Context aware answer | Pass |
6. CONCLUSION
-------------
The Multilingual Agentic RAG Chatbot successfully integrates LLMs, Vector
Databases, RAG pipeline, multilingual NLP, and agentic workflows. The system
supports natural interaction, runtime document uploads, and autonomous
generation of structured outputs like research papers.