Agentic RAG Platform

An autonomous research agent that answers questions from your documents. Built with LangGraph, Gemini, and deployed on Google Cloud Run.

Live Demo: https://agentic-rag-platform-21171800375.us-central1.run.app/docs

---

What It Does

Upload any PDF or text document, then ask questions about it. The agent retrieves the most relevant sections, generates a grounded answer, grades its own output, and retries if the quality is poor — all autonomously.

Example:

POST /ingest  →  upload company_report.pdf
POST /query   →  "What were the top reasons for customer churn in Q3?"
              ←  Answer with cited sources

---

Architecture

User
 │
 ▼
FastAPI (Cloud Run)
 │
 ├── POST /ingest ──► Chunk document ──► Vertex AI Embeddings ──► Vector Store
 │
 └── POST /query ───► LangGraph Agent
                           │
                      [retrieve] ◄──────────────────┐
                           │                        │ retry if
                      [generate]               answer poor
                           │                        │
                      [check_answer] ───────────────┘
                           │
                        answer + sources

---

Tech Stack

Layer	Technology
Agent Framework	LangGraph
LLM	Gemini 2.5 Flash (Google AI Studio)
Embeddings	Vertex AI text-embedding-004
Vector Store	LangChain InMemoryVectorStore
API	FastAPI + Uvicorn
Containerisation	Docker (two-stage build)
Deployment	Google Cloud Run

---

API Endpoints

Method	Endpoint	Description
GET	/health	Health check
GET	/docs	Interactive API playground (Swagger UI)
POST	/ingest	Upload a .pdf or .txt file
POST	/query	Ask a question, get an answer + sources

Query Example

curl -X POST https://agentic-rag-platform-21171800375.us-central1.run.app/query \
  -H "Content-Type: application/json" \
  -d '{"question": "What is LangGraph?"}'

{
  "answer": "LangGraph is a framework for building agentic workflows as state machines...",
  "sources": ["document.pdf"]
}

---

Run Locally

Prerequisites: Python 3.11+, GCP account, Google AI Studio API key

# Clone and install
git clone https://github.com/xa0412/Agentic-RAG-Platform.git
cd Agentic-RAG-Platform
python -m venv venv
source venv/Scripts/activate  # Windows
pip install -r requirements.txt

# Configure environment
cp .env.example .env
# Fill in your values in .env

# Authenticate with GCP (for embeddings)
gcloud auth application-default login

# Start the server
uvicorn api.main:app --reload --port 8080

Then open http://127.0.0.1:8080/docs

---

Deploy to Cloud Run

gcloud run deploy agentic-rag-platform \
  --source . \
  --region us-central1 \
  --platform managed \
  --allow-unauthenticated \
  --memory 1Gi \
  --max-instances 2 \
  --set-env-vars "GCP_PROJECT_ID=YOUR_PROJECT,GCP_REGION=us-central1,GEMINI_API_KEY=YOUR_KEY"

---

Environment Variables

See .env.example for the full list.

Variable	Description
GCP_PROJECT_ID	Your GCP project ID
GCP_REGION	GCP region (us-central1 recommended)
GEMINI_API_KEY	Google AI Studio API key — get one free