Chatbot Backend (RAG) – FastAPI

Production-ready backend for a Retrieval-Augmented Generation (RAG) chatbot. It handles document ingestion, chunking, embeddings, vector search, and chat generation via local LLMs (Ollama) or API-based providers.

Features

• FastAPI service with API versioning and OpenAPI docs
• RAG pipeline: document upload → chunking → embeddings → vector search → chat
• PostgreSQL + pgvector for vector search
• MinIO for object storage (documents, artifacts)
• Redis for caching/session/queues
• Celery (broker/backend configurable) for async tasks
• Ollama integration for local LLMs (model pulled on startup)
• JWT auth (access/refresh), email verification, password reset
• Nginx reverse proxy

Architecture

• API: FastAPI (app/main.py), paginated endpoints, middleware, error handling
• Auth: JWT, OAuth-style flows under /{VERSION}/oauth
• Knowledge Base: /{VERSION}/kb for KB entities
• Documents: /{VERSION}/document for upload/ingest
• Chunks: /{VERSION}/chunking for text splitting
• Embeddings: /{VERSION}/embedding to generate/store vectors
• Chat: /{VERSION}/chat for RAG conversations; /{VERSION}/c for conversations; /{VERSION}/message for messages
• API Keys: /{VERSION}/api-key

Backed services via docker-compose.yml:

• nginx on port 80 → routes to backend and services
• backend (FastAPI) on port 8000
• db PostgreSQL (pgvector) on port 5433 (host)
• minio on ports 9000/9001
• redis on port 6379
• ollama on port 11435 (mapped to container 11434)

Quick Start (Docker)

1. Copy and configure environment

cp .env.example .env
# Edit .env with your secrets and settings (see Environment Variables below)

2. Start the stack

docker compose up -d --build

3. Verify services

# Backend health
curl http://localhost:8000/{VERSION}/health

# Ollama tags (through host mapping)
curl http://localhost:11435/api/tags

# MinIO Console
open http://localhost:9001

4. API docs

• Swagger UI: http://localhost:8000/docs
• ReDoc: http://localhost:8000/redoc

If running behind nginx on port 80, the same routes are available at http://localhost.

Local Development (no Docker)

Prerequisites: Python 3.13, PostgreSQL with pgvector, Redis, MinIO, Ollama (optional if using remote LLM)

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

cp .env.example .env
# Edit .env

# Run database migrations (optional if using alembic)
alembic upgrade head

# MinIO start
minio server --console-address ":9001" /data

# Ollama start
ollama server

# Redis start
redis-server

# Celery task
celery -A app.celery_task.c_app worker -l info

# Start dev server
uvicorn app.main:app --reload --host 0.0.0.0 --port 8000

Environment Variables

The app loads settings from .env (see app/config.py). At minimum, provide:

# Database
DATABASE_URL_ASYNCPG_DRIVER=postgresql+asyncpg://USER:PASS@HOST:PORT/DB
DATABASE_URL_PSYCOPG_DRIVER=postgresql+psycopg://USER:PASS@HOST:PORT/DB
PSYCOPG_CONNECT=host=HOST port=PORT dbname=DB user=USER password=PASS

# MinIO
MINIO_URL=http://minio:9000
MINIO_ACCESS_KEY=changeme
MINIO_SECRET_KEY=changeme
BUCKET_NAME=chatbot

# Auth / Security
SECRET_KEY=changeme
SALT=changeme
ALGORITHM=HS256
ACCESS_TOKEN_EXPIRE_MINUTES=30
REFRESH_TOKEN_EXPIRE_DAYS=7
JTI_EXPIRY_SECOND=1209600

# Redis
REDIS_URL=redis://redis:6379/0

# Email (for verification/reset)
MAIL_USERNAME=your@email
MAIL_PASSWORD=app-password
MAIL_FROM=sender@email
MAIL_SERVER=smtp.example.com

# Celery
BROKER_URL=redis://redis:6379/1
BACKEND_URL=redis://redis:6379/2

# Models / LLM
EMBEDDING_MODEL=sentence-transformers/all-MiniLM-L6-v2
LLM_MODEL=deepseek-r1:8b
OLLAMA_HOST=http://ollama:11434

# App
DOMAIN_NAME=http://localhost
VERSION=v1

Note: docker-compose.yml maps Ollama to host 11435 and runs ollama pull $LLM_MODEL on startup.

Database and Migrations

• Alembic config in migrate/ and alembic.ini
• Run migrations:

alembic upgrade head

Common Endpoints

• Health: GET /{VERSION}/health
• Auth: /{VERSION}/oauth/*
• KB: /{VERSION}/kb/*
• Documents: /{VERSION}/document/*
• Chunking: /{VERSION}/chunking/*
• Embedding: /{VERSION}/embedding/*
• Chat: /{VERSION}/chat/*
• Conversation: /{VERSION}/c/*
• Message: /{VERSION}/message/*
• API Key: /{VERSION}/api-key/*

Actual schemas and request/response bodies are documented in Swagger.

Development Notes

• App entry: app/main.py
• Config: app/config.py
• Services and routers organized by domain under app/
• File uploads: uploads/
• Nginx config: nginx.conf
• Docker build: Dockerfile
• Compose stack: docker-compose.yml

Testing

You can add tests under app/tests/ (folder present). Run with:

pytest -q

License

MIT – see LICENSE.