fastapi-template

FastAPI Template with Industry Standard Observability

Hatchet

Background workflows are powered by Hatchet.

Required env:

• HATCHET_CLIENT_TOKEN

Run workers:

# General worker
cd api-workers-general && make run-worker

# ML worker
cd api-workers-ml && make run-worker

OpenTelemetry

You can choose where to send OpenTelemetry traces. You can use any OpenTelemetry compatible backend, but the template has built-in support for Logfire and Langfuse.

Logfire

• Logfire is Uncomplicated Observability for Python by Pydantic Team.
• Create a new project in Logfire with name fastapi-template

Set in .env

OLTP_LOG_METHOD=logfire
LOGFIRE_TOKEN=FILL_IT

Langfuse

Set in .env:

OLTP_LOG_METHOD=langfuse
LANGFUSE_BASE_URL=FILL_IT
LANGFUSE_PUBLIC_KEY=FILL_IT
LANGFUSE_SECRET_KEY=FILL_IT

Migrations

alembic upgrade "<revision_id>"
alembic upgrade "head"

Reverting migrations

If you want to revert migrations, you should run:

# revert all migrations up to: revision_id.
alembic downgrade <revision_id>

# Revert everything.
alembic downgrade base

Migration generation

To generate migrations you should run:

# For automatic change detection.
alembic revision --autogenerate

# For empty file generation.
alembic revision

Example API Requests

General workflow request:

TASK_ID=$(curl -X POST http://localhost:8000/api/v1/tasks/general/long-running \
  -H "Content-Type: application/json" \
  -d '{"duration": 30}' \
  -s | jq -r '.metadata.id')

curl -X GET http://localhost:8000/api/v1/tasks/general/$TASK_ID

ML requests:

curl -X POST http://localhost:8000/api/v1/tasks/ml/inference \
  -H "Content-Type: application/json" \
  -d '{"model_id": "test-model", "input_data": {"features": [1, 2, 3]}}'

curl -X POST http://localhost:8000/api/v1/tasks/ml/training \
  -H "Content-Type: application/json" \
  -d '{"dataset_id": "dataset-123", "model_configuration": {"input_size": 10, "output_size": 1}, "hyperparameters": {"epochs": 5, "learning_rate": 0.01, "batch_size": 32}}'