AI Inference API

An asynchronous REST API built with FastAPI that validates, routes and stores text-inference requests in PostgreSQL.

What it does

• Accepts text and a task name (sentiment or toxicity) at POST /predict
• Validates the payload with Pydantic (empty or invalid input is rejected with HTTP 422)
• Routes the text to the matching model function
• Saves each result in PostgreSQL through SQLAlchemy (async), committing in one transaction and rolling back on failure
• Returns the result, which can be fetched later at GET /results/{id}

The two models are simple word-based rules, so the focus is on the API and data layer. The routing table in inference.py makes it easy to plug in a trained model.

Run it

python -m venv .venv
.venv\Scripts\activate          # Mac/Linux: source .venv/bin/activate
pip install -r requirements.txt
docker compose up -d db
uvicorn main:app --reload

Open http://127.0.0.1:8000/docs to try it.

Endpoints

Method	Path	Purpose
POST	/predict	Validate, route, run and save a prediction
GET	/results/{id}	Fetch a saved result
GET	/health	Basic health check

Roadmap

Automated tests, Docker for the API, load testing, a trained ML model, and migrations.