Skip to content

Full Platform Setup

This guide walks you through deploying the complete Skyulf platform — backend API, Celery workers, Redis, and the web UI.

Just want the Python library? See Installation to pip install skyulf-core and skip this page.

Prerequisites

Requirement Notes
Docker Desktop (recommended) Docker Compose launches everything in containers — no host Python or Redis needed.
Python 3.12 (manual path) Only if you prefer running services directly on your machine.
Redis (manual path) Required for Celery background tasks. Install natively, via WSL, or run docker run -d -p 6379:6379 redis:7.
Git To clone the repository.

1. Clone the Repository

git clone https://github.com/flyingriverhorse/Skyulf.git
cd Skyulf

2. Configure Environment Variables

Copy the example environment file and review the settings:

cp .env.example .env

Key variables in .env:

Variable Default Description
DB_TYPE sqlite Database backend (sqlite or postgres)
USE_CELERY false Set true to enable Redis-backed background jobs (requires Redis)
CELERY_BROKER_URL redis://127.0.0.1:6379/0 Redis URL for Celery message broker
REDIS_URL redis://127.0.0.1:6379/0 Redis URL for caching
SECRET_KEY (random, regenerated each restart) Required in production — set a stable value or JWT tokens will break on restart
DEBUG true Set false in production

Minimal setup: For local development with SQLite and no Celery (the default), no changes are needed — just cp .env.example .env and start the server.

From the repository root:

docker compose up --build

This starts three containers:

Service Port Purpose
api 8000 FastAPI backend + static frontend
redis 6379 Message broker for Celery
worker — Celery worker for background training jobs

Verify it's running:

curl http://127.0.0.1:8000/health
# Expected: {"status": "healthy", ...}

Useful commands:

# Run in background
docker compose up --build -d

# Follow API logs
docker compose logs -f api

# Stop everything
docker compose down

Tip: Docker Compose mounts your working tree, so local code edits are picked up on reload.

3B. Manual Setup (No Docker)

Create a virtual environment

python -m venv .venv

# Windows PowerShell
.\.venv\Scripts\Activate.ps1

# macOS / Linux
source .venv/bin/activate

Install dependencies

pip install --upgrade pip
pip install -r requirements-fastapi.txt

Start Redis (if USE_CELERY=true)

# Option 1: Docker (simplest)
docker run --name skyulf-redis -p 6379:6379 -d redis:7

# Option 2: Native install (Linux)
sudo apt install redis-server && sudo systemctl start redis

# Option 3: Skip Redis entirely
# Set USE_CELERY=false in your .env file

Launch the API server

python run_skyulf.py

Or with uvicorn directly:

python -m uvicorn backend.main:app --reload --host 127.0.0.1 --port 8000

Start the Celery worker (optional)

In a second terminal with the virtualenv activated:

python -m celery -A celery_worker.celery_app worker --pool=solo --loglevel=info --queues mlops-training

4. Verify the Stack

Once the server is running, check these URLs:

URL What it is
http://127.0.0.1:8000 Web dashboard (main UI)
http://127.0.0.1:8000/docs Swagger UI — interactive API explorer
http://127.0.0.1:8000/redoc ReDoc — API reference documentation
http://127.0.0.1:8000/health Health check endpoint
http://127.0.0.1:8000/health/detailed Detailed diagnostics (DB connectivity, Redis/Celery availability, Snowflake status)

5. First Login

Authentication is not yet enabled in this release — the platform is accessible without credentials. When auth is implemented, credentials will be configured via environment variables (AUTH_FALLBACK_* settings in .env).

Heads-up: Do not expose the API on a public network until authentication is implemented.

6. First Steps After Setup

  1. Open the dashboard at http://127.0.0.1:8000.
  2. Upload data: Navigate to Data Sources and upload a CSV file.
  3. Explore your data: Open EDA to see automated profiling, distributions, and outlier alerts.
  4. Build a pipeline: Open the ML Canvas, drag nodes, connect them, and hit Execute.
  5. Deploy a model: Register a trained model and test it from the Deployments page.

For a detailed step-by-step walkthrough, see Platform Walkthrough.

Troubleshooting

Problem Fix
Connection refused on port 8000 Is the API running? Check docker compose logs api or your terminal.
Celery tasks stuck in "pending" Verify Redis is running: redis-cli ping should return PONG.
ModuleNotFoundError on startup Activate your virtualenv and run pip install -r requirements-fastapi.txt.
Database errors after update Delete mlops_database.db (SQLite) to start fresh — it auto-recreates on boot.
CORS errors in browser Add your frontend URL to CORS_ORIGINS in .env. Default allows localhost:3000 and localhost:8080.

For more issues, see Troubleshooting.