🔍 GenAI Fundamental Industry Analysis (Local RAG Pipeline)

A lightweight, locally deployable GenAI pipeline that performs fundamental analysis of industries by analyzing PDFs, Excel sheets, and API sources assigned by the user. This project leverages Retrieval-Augmented Generation (RAG) with local LLMs and embedding models.

---

🧠 Features

• Ingest data from PDF, Excel, and API sources
• Preprocess and store data as clean text files
• Generate vector embeddings locally with sentence-transformers
• Use lightweight LLMs such as phi-3-mini or mistral-7b
• Perform contextual question-answering via RAG
• Simple CLI or Streamlit interface for interaction

---

🧱 Project Structure

genai\_fundamental\_analysis/
├── data\_sources/        # Raw input data (PDF, Excel, API)
├── configs/             # Configuration YAMLs
├── embeddings/          # Vector store & intermediate text chunks
├── models/              # Local LLMs if needed
├── scripts/             # Modular pipeline scripts
├── main.py              # Entry point script
└── README.md            # Project documentation

---

⚙️ Setup Instructions

# 1. Clone the repository
git clone https://github.com/{os.userid}/genai_fundamental_analysis.git
cd genai_fundamental_analysis

# 2. Create virtual environment (optional but recommended)
python -m venv venv
source venv/bin/activate  # or venv\Scripts\activate on Windows

# 3. Install dependencies
pip install -r requirements.txt

# 4. Run the pipeline
python main.py

---

🛠️ Tools & Frameworks

• LLMs: phi-3-mini / mistral-7b (via llama.cpp)
• Embeddings: BAAI/bge-small-en (sentence-transformers)
• Vector DB: FAISS / Chroma
• Frameworks: LangChain or LlamaIndex for RAG
• Interface: Streamlit (optional UI)

---

✅ Use Case Example

“Analyze all uploaded PDFs and Excel files related to the Pharma sector. Summarize key risks, growth indicators, and recent developments.”

The system will:

1. Ingest and parse source files
2. Create vector embeddings
3. Use a local LLM to search and synthesize answers
4. Present results in CLI or Streamlit

---

🤝 Contributing

Contributions are welcome! Feel free to fork the repo, raise issues, or submit pull requests.

You can also connect with me on LinkedIn to collaborate on similar use cases.

---

📄 License

This project is licensed under the MIT License.

---

🙌 Acknowledgements

• LangChain
• HuggingFace Transformers
• llama.cpp
• BAAI bge-small-en

---