From ML Beginner to Production Engineer: How I’m Leveling Up My AI Projects

October 12, 2025

This content originally appeared on DEV Community and was authored by marcusmayo

From training toy models to shipping real ML systems — here’s what that journey really looks like.

Most people start their ML learning journey in Jupyter notebooks. But when you want your model to serve real users, things get serious — and a lot more complex.

Here’s how the levels break down

Level 1 – Learning the Basics

Clean datasets (Kaggle, UCI)
Jupyter notebooks & visualization
Simple metrics and evaluation

Level 2 – Professional Data Science

Handling messy, real-world data
Organized code + config files
Feature engineering & tuning
Git for reproducibility

Level 3 – Machine Learning Engineering

Containerized model APIs (Docker/FastAPI)
MLflow for tracking + model registry
CI/CD pipelines
Monitoring & scaling on AWS/GCP

I’m documenting my path across these levels — moving from education to execution.

The next phase: Level 4, where models scale, retrain automatically, and support real users.