On my first SAP team, I built a retrieval-augmented chatbot over our product documentation - well before RAG was a common term. It let customers chat with the docs to get answers fast.
Highlights
- Implemented a simple RAG pipeline using compressed vectors and word2vec embeddings.
- Ranked answers with a weighted, vector-based cosine-similarity model over the documentation.
- Deployed on the product documentation page so customers could chat directly with the docs.
Stack
PythonWord2VecNLTKNLP
Python