A Retrieval-Augmented Generation (RAG) system that enables users to query company earnings data using natural language.
The application allows users to ask questions such as revenue, earnings growth, or company performance using conversational queries instead of manually searching through financial reports.
It combines vector search, structured filtering, and LLM reasoning to generate grounded answers from financial data.
The interface allows users to:
- Select company ticker
- Choose earnings quarter
- Ask financial questions
- Retrieve grounded responses based on earnings reports
Example query:
what is the revenue?
Example response:
The revenue reported in Nvidia's Q3 2024 earnings call was $18.1 billion.
The system follows a Retrieval-Augmented Generation pipeline:
- User Query
User submits a natural language financial question.
- Frontend Interface
Web application allows users to select:
- company ticker
- quarter
- year
- query
- FastAPI Backend
The backend processes the request and coordinates retrieval and generation.
- Metadata Filtering
The system filters financial data using:
- company ticker
- quarter
- year
- Context Retrieval
Relevant financial content is retrieved using vector search from the financial dataset.
- Prompt Augmentation
Retrieved context is combined with the user query.
- LLM Response Generation
The language model generates an answer grounded in the retrieved financial context.
- Response Returned
The final answer is returned to the user interface.
Financial reports and earnings transcripts contain valuable insights but are often long and difficult to navigate.
This project demonstrates how modern AI systems can transform financial data into a conversational interface that allows users to explore earnings information quickly and efficiently.
It showcases an applied AI workflow combining:
- backend engineering
- retrieval systems
- natural language interfaces
- LLM reasoning
Natural language querying of financial data Retrieval-Augmented Generation architecture Quarter and company filtering Vector search for contextual retrieval Grounded responses from financial reports FastAPI backend API Interactive web interface
Users can ask questions such as:
What is the revenue for this quarter?
What is the profit growth?
What drove Nvidia's revenue growth?
What is the gaming revenue?
What is the earnings outlook?
Python FastAPI Vector Search / RAG Pipeline Large Language Models (LLMs) HTML / JavaScript frontend Financial data processing
AI-Driven-Stock-Market-Querying/
│
├── app.py
├── retriever.py
├── llm_service.py
├── database.py
├── requirements.txt
│
├── static/
│ └── index.html
│
├── data/
│ ├── earnings_transcripts
│ └── processed_financial_data
│
├── demo.png
├── architecture.png
Clone the repository
git clone https://github.com/MuntahaShams/AI-Driven-Stock-Market-Querying.git
cd AI-Driven-Stock-Market-Querying
Install dependencies
pip install -r requirements.txt
Start the FastAPI server
uvicorn app:app --reload
Open the application in your browser
http://localhost:8000
Financial documents present several challenges:
- long earnings transcripts
- noisy textual data
- multiple financial metrics
- different formats across companies
This project addresses these challenges using retrieval-based context selection combined with LLM reasoning.
Add real-time financial APIs Add financial chart visualization Support multiple company comparison Improve context ranking and retrieval Add evaluation metrics for RAG performance
Muntaha Shams AI Engineer — LLMs | NLP | Computer Vision | Document AI
GitHub https://github.com/MuntahaShams
Portfolio https://muntahashams.github.io/portfolio/projects