📚 Internal Docs Q&A Dashboard — Powered by LangChain RAG

RAG-Assistant is a modern web dashboard for secure, AI-powered search and Q&A over your organization’s internal documents. Easily upload PDFs or TXT files, ask natural language questions, and receive context-rich answers with transparent source attribution. Includes analytics for usage, accuracy, and performance monitoring.

---

✨ Key Features

• 📄 Document Upload: Build a custom knowledge base from internal PDFs and text files.
• 💬 Natural Language Q&A: Ask questions and get answers sourced directly from your docs.
• ✈️ Aviation Data RAG: Integrate and query 80+ aviation data reports as a specialized vector database for aviation-related Q&A, enabling deep domain insights and analysis.
• 🔄 Multi-turn Conversations: Chat with context-aware memory for richer discussions.
• 🔎 Source Attribution: Every answer lists the document sources and page numbers.
• 📊 Analytics Dashboard: Track query count, response latency, and answer accuracy.
• 🤖 Multi-Model Support: Choose from leading OpenAI and Groq models.
• 🛡️ Session Management: Persistent chat history by user/session.
• 🛠️ LangSmith Monitoring: Integrated tracing for debugging and performance insights.

---

🏢 Typical Use Cases

• Internal Knowledge Base: Empower staff to search policies, manuals, HR docs, and more.
• Support Portal: Fast answers for IT, HR, or helpdesk teams based on company docs.
• Research & Analysis: Upload reports, ask complex questions, and get cited answers.
• Learning Platform: Deep-dive discussions with memory of previous context.
• Aviation Data Search: Analyze and explore aviation datasets for research, compliance, or operational insights.

---

🚀 Quick Start

1. Clone the Repository

   git clone https://github.com/JSM2512/RAG-Assistant.git
   cd RAG-Assistant

2. Set Up Python Environment

   conda create -p venvlangchain python=3.12
   conda activate venvlangchain

3. Install Dependencies

   pip install -r requirements.txt

4. Add API Keys Create a .env file in the repo root:

   OPENAI_API_KEY=your_openai_api_key
   GROQ_API_KEY=your_groq_api_key
   LANGCHAIN_API_KEY=your_langsmith_api_key
   LANGCHAIN_PROJECT=LangchainRAG
   LANGCHAIN_TRACING_V2=true

5. Launch the Dashboard

   streamlit run app.py

---

🖥️ Dashboard Overview

• Chat Tab: Upload documents, select your preferred model, and ask questions — see answers with sources.
• Analytics Tab: Visualize total queries, latency, and answer accuracy over time.
• Session Selector: Each user or session gets its own persistent chat history.
• Aviation Data RAG Tab: Instantly query 80+ aviation datasets using RAG, suitable for aviation ops, research, and compliance.

---

🏗️ Architecture

┌───────────────┐   ┌─────────────┐   ┌───────────────┐
│ Streamlit UI  │──▶│ RAG Engine  │──▶│ Vector Store  │
└───────────────┘   └─────────────┘   └───────────────┘
        │                │                 │
        ▼                ▼                 ▼
 ┌─────────────┐   ┌─────────────┐   ┌──────────────┐
 │ User Upload │   │ LLM Models  │   │ Analytics DB │
 └─────────────┘   └─────────────┘   └──────────────┘
        │
        ▼
┌───────────────────────┐
│ Aviation Data VectorDB│
└───────────────────────┘

---

🤖 Supported Models

Provider	Model Name	Model ID
OpenAI	GPT-4o, GPT-4-turbo, GPT-4	gpt-4o, gpt-4-turbo, gpt-4
Groq	Deepseek LLaMA3, Gemma, Mistral	Deepseek-R1-Distill-Llama-70b, Gemma2-9b-It, Mistral-Saba-24b

---

✈️ Aviation Data RAG

• Integrated 80+ aviation datasets as a vector database for domain-specific retrieval.
• Use Cases: Safety analysis, fleet management, regulatory compliance, route optimization, research.
• Query Examples: "Show all incidents for Boeing 737 in 2023", "List ICAO codes for European airports", "Analyze global flight delays trends."
• Source Attribution: Answers cite the aviation dataset(s) used.

---

📊 Analytics Dashboard

• Query Tracking: Count of questions asked per session.
• Latency Monitoring: Average and per-query response time (sec).
• Accuracy Feedback: Track user feedback on answer correctness.
• Interactive Charts: Visualize latency and accuracy trends.

---

📁 Project Structure

RAG-Assistant/
├── app.py             # Main Streamlit dashboard
├── Project.ipynb      # Core RAG prototype notebook
├── requirements.txt   # Python dependencies
├── .env               # API keys/config
├── aviation_data/     # Aviation datasets (vector db source)
└── README.md          # This documentation

---

🔒 Security & Privacy

• Local Processing: All uploaded docs are processed locally and not sent to third-party servers (except for LLM query).
• Session Isolation: Each chat session is isolated for privacy.
• API Keys: Store keys securely in .env file.
• Aviation Data: Aviation datasets processed locally for compliance and privacy.

---

🎥 Demo Workflow

1. Upload internal docs (PDF/TXT)
2. Ask questions about those docs or aviation datasets
3. Get answers with source citation
4. Review analytics for usage and accuracy

---

🧠 How It Works

1. Document Ingestion: Upload, chunk, and embed docs into a vector store.
2. Aviation Data Load: Aviation datasets embedded as a specialized vector store for domain queries.
3. RAG Pipeline: Retrieve relevant chunks using ensemble search (semantic + keyword).
4. Chat Engine: Generate contextual answers with selected LLM, citing sources.
5. Analytics: Monitor performance and collect feedback for continuous improvement.

---

💡 Tips

• Use specific session IDs for different teams/users.
• Upload up-to-date docs for the best answers.
• Review sources for traceability.
• Share the dashboard with your team for internal knowledge sharing.
• Leverage Aviation Data RAG for deep aviation insights and analysis.

---