dev-archive

RAG pipeline for querying and analyzing software engineering case studies. Locally hosted with a FastAPI backend, React frontend, Pinecone vector DB, and SQLite for chat history.

---

Prerequisites

• Python + Poetry
• Node.js + npm
• Pinecone account
• OpenAI API key

Environment Variables

Create a .env file in the project root:

OPENAI_API_KEY=[insert key]
PINECONE_API_KEY=[insert key]
PINECONE_INDEX_NAME=dev-archive
PINECONE_INDEX_NAMESPACE=case-studies

---

Server (FastAPI · port 8000)

cd server
pipx install poetry        # skip if already installed
poetry config virtualenvs.in-project true
poetry install
eval $(poetry env activate)
fastapi dev

Client (React · port 5173)

cd client
npm install
npm run dev

Open http://localhost:5173 — API requests proxy automatically to port 8000.

---

Pinecone Setup

1. Create a Pinecone index named dev-archive
2. Create a namespace case-studies
3. Run the ingestion script to embed and upsert chunks:

cd server
eval $(poetry env activate)
python ingestion/embed_upsert.py   # reads from ingestion/chunks.json

---

Storage

Store	Details
Pinecone	Vector embeddings for semantic search (llama-text-embed-v2, top-10 retrieval)
SQLite	Chat thread history saved to server/history/chat_history.db