Getting Started Tutorial

Transform a simple machine learning function into a production-ready pipeline, solving real challenges along the way.

What You'll Build

By the end of this tutorial, you'll have:

• ✅ Reproducible ML pipeline: Automatic tracking of all runs and results
• ✅ Configurable experiments: Change parameters without touching code
• ✅ Multi-step workflow: Data loading → preprocessing → training → evaluation
• ✅ Large dataset handling: Efficient storage and retrieval of data artifacts
• ✅ Shareable results: Model artifacts and metrics that persist between runs
• ✅ Deployment ready: Same pipeline runs on laptop, containers, or Kubernetes

The Journey

Each chapter tackles a real problem you'll face moving from "works on my laptop" to production:

1. The Starting Point - A typical ML function with common problems
2. Making It Reproducible - Track everything automatically
3. Adding Flexibility - Configure without code changes
4. Connecting the Workflow - Multi-step ML pipeline
5. Handling Large Datasets - Efficient data management
6. Sharing Results - Persistent model artifacts and metrics
7. Running Anywhere - Same code, different environments

Prerequisites

• Basic Python knowledge
• Familiarity with scikit-learn (we'll use simple examples)
• Python environment with runnable installed: pip install runnable[examples]

Time Investment: ~30-45 minutes total, designed for step-by-step learning

---

Ready to start? → The Starting Point