Apppublic

mahabub-unlocked/autonomous-research-assistant

sourceHugging Facemitupdated 6mo agoView on Hugging Face

0likes

App README

🧠 Autonomous Research Assistant

<br>

![Live Demo](https://huggingface.co/spaces/mahabub-unlocked/QueryMind-AI) ![LinkedIn](https://www.linkedin.com/in/md-mahabub-alam-bishal/)

<br>

একটা intelligent AI research assistant যা যেকোনো বিষয়ে real-time web search, Wikipedia analysis, এবং document processing করে বিস্তারিত উত্তর প্রদান করে — LangGraph এর ReAct agent framework দিয়ে।

</div>

🎯 Project Overview

Autonomous Research Assistant হল একটি AI-powered research assistant যা তিনটা powerful tool ব্যবহার করে —

Web Search — DuckDuckGo দিয়ে real-time internet search Wikipedia Analysis — যেকোনো বিষয়ে Wikipedia থেকে বিস্তারিত তথ্য Document Processing — PDF ও TXT file upload করে analyze করা

🚀 Live Demo

👉 [App Open করো](https://huggingface.co/spaces/mahabub-unlocked/QueryMind-AI)

Feature	Description
🔍 Web Search	Real-time internet search
📖 Wikipedia	বিস্তারিত research
📄 File Analysis	PDF ও TXT analyze
🌐 Multilingual	Bengali ও English support
⚡ Streaming	Real-time typing effect

🏗️ Architecture

User Query (Bengali / English)
        │
        ▼
┌───────────────────┐
│   ReAct Agent     │ ── Tool selection decision
│   (LangGraph)     │
└────────┬──────────┘
         │
    ┌────┴────┐
    ▼         ▼         ▼
┌────────┐ ┌────────┐ ┌────────┐
│  Web   │ │ Wiki   │ │  File  │
│ Search │ │ Search │ │ Loader │
│  🔍   │ │  📖   │ │  📄   │
└───┬────┘ └───┬────┘ └───┬────┘
    └──────────┴──────────┘
                │
                ▼
┌───────────────────────┐
│   LLaMA 3.3 70B       │ ── Response generate
│   (via Groq)          │
└───────────┬───────────┘
            │
            ▼
┌───────────────────────┐
│   Streamlit UI        │ ── Streaming output
│   Chat Interface      │ ── History + Export
└───────────────────────┘

✨ Features

Feature	Description
🔍 Web Search	DuckDuckGo দিয়ে real-time internet search
📖 Wikipedia Search	যেকোনো বিষয়ে Wikipedia থেকে বিস্তারিত তথ্য
📄 File Analysis	PDF ও TXT file upload করে analyze করা
🌐 Multi-language	Bengali 🇧🇩 ও English 🇬🇧 response support
⚡ Streaming Output	Real-time টাইপিং effect এ response
💾 Chat History	Conversation save ও load করার সুবিধা
📥 PDF Export	পুরো chat PDF এ export করা
📋 Copy Button	যেকোনো response এক click এ copy
🧠 ReAct Agent	Intelligent tool selection ও decision making

🛠️ Tech Stack

Component	Technology
Frontend	Streamlit
AI Agent	LangGraph (ReAct)
LLM	LLaMA 3.3 70B via Groq
Tools	DuckDuckGo + Wikipedia
Framework	LangChain
Language	Python 3.10+
Deployment	Hugging Face Spaces

📁 Project Structure

Autonomous Research Assistant/
│
├── 📄 app.py               # Streamlit UI & main application
├── 🤖 agent.py             # LangGraph ReAct agent setup
├── 🔧 tools.py             # Custom LangChain tools
├── 📋 requirements.txt     # Python dependencies
├── 🔒 .env                 # API keys (local only)
├── 🚫 .gitignore           # Git ignore rules
└── 📖 README.md            # Project documentation

🚀 Getting Started

Prerequisites

Python 3.10+
Groq API Key → groq.com (Free)

Installation

1. Clone the repository

bash

git clone https://github.com/mdmahabubalambishas/Autonomous Research Assistant.git
cd Autonomous Research Assistant

2. Create virtual environment

bash

python -m venv venv
venv\Scripts\activate        # Windows
source venv/bin/activate     # Mac/Linux

3. Install dependencies

bash

pip install -r requirements.txt

4. Set up environment variables

.env file বানাও project root এ —

env

GROQ_API_KEY=your_groq_api_key_here

5. Run the app

bash

streamlit run app.py

Open http://localhost:8501 in your browser.

🧪 Example Queries

🔍 Web Search

"LangGraph কী এবং কীভাবে কাজ করে?"
"২০২৫ সালে AI এর নতুন developments কী?"
"Latest news about OpenAI?"

📖 Wikipedia Research

"Artificial Intelligence এর ইতিহাস বিস্তারিত বলো"
"Quantum Computing কীভাবে কাজ করে?"
"Machine Learning vs Deep Learning পার্থক্য?"

📄 Document Analysis

১. Sidebar এ PDF upload করো
২. "এই document এর মূল বিষয় কী?" — জিজ্ঞেস করো
৩. "Summarize this document in Bengali"

🔍 How ReAct Agent Works

Step 1 → Thought
         Agent query analyze করে
         কোন tool দরকার সেটা decide করে

Step 2 → Action
         সঠিক tool select করে
         Web Search / Wikipedia / File Loader

Step 3 → Observation
         Tool এর result দেখে
         Answer sufficient কিনা check করে

Step 4 → Final Answer
         LLaMA 3.3 70B দিয়ে response generate করে
         Bengali বা English এ stream করে দেয়

⚙️ Configuration

Model পরিবর্তন (agent.py) —

python

llm = ChatGroq(
    model="llama-3.3-70b-versatile",  # LLM model
    temperature=0.2,                   # Creativity (0-1)
    max_tokens=4096                    # Response length
)

Custom Tool যোগ করা (tools.py) —

python

@tool
def my_custom_tool(query: str) -> str:
    """Tool description."""
    # Your logic here
    return result

all_tools = [search_tool, wikipedia_search, my_custom_tool]

📈 Performance

Metric	Value
Web Search response time	~2-3 seconds
Wikipedia response time	~1-2 seconds
PDF analysis time	~3-5 seconds
Languages supported	Bengali, English
Max file size	10MB
LLM model	LLaMA 3.3 70B

💡 What I Learned

✅ LangGraph ReAct Agent
   Multi-step reasoning
   Intelligent tool selection
   State management

✅ Tool Integration
   DuckDuckGo web search
   Wikipedia API
   PDF/TXT file processing

✅ Streaming Response
   Real-time output
   Typing effect implementation

✅ Multilingual LLM
   Bengali language support
   Language detection
   Response formatting

✅ Chat Management
   Conversation history
   PDF export
   Copy functionality

🤝 Contributing

bash

# Fork করো → Clone করো → Branch তৈরি করো
git checkout -b feature/AmazingFeature

# Changes করো → Commit করো
git commit -m "Add AmazingFeature"

# Push করো → Pull Request খোলো
git push origin feature/AmazingFeature

📄 License

This project is licensed under the MIT License.

👨‍💻 Author

Md Mahabub Alam Bishal AI/ML Engineer | LLM & Generative AI Enthusiast

![LinkedIn](https://www.linkedin.com/in/md-mahabub-alam-bishal/) ![GitHub](https://github.com/mdmahabubalambishal) ![HuggingFace](https://huggingface.co/mahabub-unlocked) ![Email](mailto:mdmahabubalambishal@gmail.com)

</div>

🙏 Acknowledgements

—LangChain — LLM framework
—LangGraph — Agent orchestration
—Groq — Fast LLM inference
—DuckDuckGo — Privacy-first web search
—Wikipedia — Knowledge base
—Streamlit — UI framework
—Hugging Face — Deployment platform

⭐ এই project টা useful লাগলে GitHub এ Star দাও!

Made with ❤️ by Mahabub

Live Demo: huggingface.co/spaces/mahabub-unlocked/QueryMind-AI GitHub: github.com/mdmahabubalambishal/QueryMind-AI

</div>