Portfolio Chatbot — Backend

FastAPI backend for the AI chatbot on tiago-coutinho.com. Answers questions about Tiago Coutinho using a system prompt built from his biography. Responses are streamed in real time.

Features

• Streaming responses via SSE
• Rate limiting per IP (5 requests/minute by default)
• Conversation history managed by the frontend — no server-side persistence
• Switchable LLM provider per request (OpenAI or local model)
• System prompt loaded from app/about-me.md at startup

LLM Providers

Provider	Status	Details
OpenAI (gpt-4o-mini)	✅ Available	Used by default
Local model	🚧 Not yet implemented	Planned: 1B–3B parameter model self-hosted on Raspberry Pi 5 with RAG

API

POST /chat

{
  "messages": [
    { "role": "user", "content": "Who is Tiago?" },
    { "role": "assistant", "content": "Tiago is..." },
    { "role": "user", "content": "What does he work on?" }
  ],
  "provider": "openai"
}

Field	Type	Required	Default	Description
messages	array	yes	—	Full conversation history
messages[].role	string	yes	—	"user" or "assistant"
messages[].content	string	yes	—	Message text
provider	string	no	"openai"	"openai" or "local"

Returns a text/event-stream response with text chunks. Returns 429 when the rate limit is exceeded, with a JSON body indicating when to retry.

GET /health

Returns {"status": "ok"}. Used by Docker and Nginx to verify the service is running.

Running Locally

cp .env.example .env
# fill in your OpenAI API key

uv run uvicorn app.main:app --reload

Or with Docker:

docker build -t portfolio-chatbot:latest .
docker run -d --name portfolio-chatbot -p 8000:8000 --env-file .env portfolio-chatbot:latest

Configuration

All settings are loaded from a .env file. See .env.example for the full list.

Variable	Default	Description
OPENAI_API_KEY	—	OpenAI API key (required)
OPENAI_MODEL	gpt-4o-mini	OpenAI model to use
OPENAI_TEMPERATURE	0.5	Response creativity (0.0–2.0)
OPENAI_MAX_TOKENS	500	Maximum tokens per response
RATE_LIMIT	5/minute	Rate limit per IP
ALLOWED_ORIGINS	["https://tiago-coutinho.com"]	CORS allowed origins

Project Structure

app/
├── main.py               # FastAPI app, CORS, rate limiting
├── config.py             # Settings and shared instances
├── about-me.md           # System prompt content
├── routers/
│   └── chat.py           # /chat and /health endpoints
├── schemas/
│   └── chat.py           # Request/response models
└── services/
    ├── base.py            # LLMService abstract class
    ├── openai_service.py  # OpenAI implementation
    └── local_service.py   # Local model (not yet implemented)