Michelangelo User Guides
This guide provides you step by step how to build, train, and deploy machine learning models at scale using Michelangelo's unified ML platform.
Getting Started: Your First ML Workflow
New to Michelangelo? Follow this step-by-step path to build your first end-to-end ML pipeline:
1. Understand ML Pipelines (Start Here)
Before diving into specific tasks, understand how Michelangelo orchestrates ML workflows:
- ML Pipelines Overview - Learn about tasks, workflows, and pipelines
- Getting Started with Pipelines - Build your first pipeline in 30 minutes
2. Prepare Your Data
Get your data ready for training:
- Prepare Your Data - Load, clean, and split datasets using Ray and Spark
3. Train Your Model
Develop and train your ML model:
- Train and Register a Model - Train locally or at scale with distributed computing
4. Manage Your Models
Version and organize trained models:
- Model Registry Guide - Version, track, and manage trained models
5. Deploy Your Model
Serve predictions in production:
- Deploy Models - Deploy models for real-time inference and batch scoring
Quick Navigation
Core Concepts
| Concept | Learn About It |
|---|---|
| How to build a pipeline | ML Pipelines Overview |
| When to use each execution mode | Pipeline Running Modes |
| How caching and resumability work | Caching and Resume |
| How to iterate rapidly | File Sync Testing |
Specific Tasks
Advanced Topics
- Set Up Pipeline Triggers - Schedule pipelines to run automatically
- CLI Reference - Command-line tools for pipeline and project management
- Project Management - Create and manage MA Studio projects
Learning by Examples
Choose a tutorial based on your ML domain:
Traditional Machine Learning
| Example | Description | Techniques |
|---|---|---|
| Boston Housing Regression | Predict house prices using tabular data with XGBoost | Feature engineering, distributed training |
Natural Language Processing
| Example | Description | Techniques |
|---|---|---|
| BERT Text Classification | Classify text using pre-trained transformer models | Fine-tuning, distributed GPU training |
| GPT Fine-tuning | Train large language models with LoRA adapters | Memory optimization, multi-GPU scaling |
Recommendation Systems
| Example | Description | Techniques |
|---|---|---|
| Amazon Books Recommendation | Build dual-encoder recommendation system | Embedding learning, similarity search |
What's next?
- New to ML Pipelines? Start with the ML Pipelines Overview, then follow Getting Started with Pipelines
- Have your pipeline running? Learn about Pipeline Running Modes and Caching and Resume
- Ready for production? Set up Pipeline Triggers for automated scheduling