AI / Data / Retrieval

← Work index

Stock / RAG System

Groq-powered RAG over 60 major Indian stocks across 9 sectors — real-time Yahoo Finance data, Streamlit UI.

Build
R&D experiment
Year
2025
Status
experiment

Problem

Financial questions deserve answers grounded in real data — not generated confidence. This experiment builds the pipeline that makes grounding possible.

Approach

  • Groq-powered RAG pipeline over 60 major Indian stocks across 9 sectors
  • Real-time Yahoo Finance data ingested into retrievable chunks
  • Streamlit interface with dark and light themes
  • Evaluation loops checking that answers trace back to source data

Retrieval decides everything

Generation is the easy part. The work is ingestion, chunking, and evaluation — iterating until the retrieved context is the right context.

Outcome

Experiment — architecture documented, no performance claims.

Stack

Groq RAG / Streamlit / Yahoo Finance / Embeddings / Evaluation

Proves — AI development →

Next case study — In progress

Majdoor Haq