Back to work

RFP Auto-Responder

RAG system that drafts responses to RFP documents from a private knowledge base.

Solo Developer · 2024

The Problem

Responding to RFPs means answering the same kinds of questions repeatedly, scattered across past proposals and documentation.

What I Built

A Streamlit app that embeds a private corpus into ChromaDB, retrieves the most relevant passages for each RFP question, and uses Google Gemini to draft grounded responses the user can edit.

Key Engineering Decisions

  • Local vector store (ChromaDB) for a fast, dependency-light prototype
  • Retrieval-grounded generation to keep answers tied to real source material
  • Streamlit for a shippable interface in days, not weeks

Stack

PythonChromaDBGoogle GeminiStreamlit