Keyboard shortcuts

Press or to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Month 6 — MLOps and Production

🎯 Goal of this month

**This month is the most important for your main goal.**To become an ML Engineer / MLOps Engineer, the knowledge from this month becomes the center of your portfolio.

By the end of the month you will be able to:

  • Know the MLOps lifecycle from start to finish
  • Track experiments and version models with MLflow
  • Ensure data versioning and reproducibility with DVC
  • Serve ML models via FastAPI + BentoML/TorchServe
  • ML deployment on Docker + Kubernetes
  • Monitoring and drift detection with Prometheus + Grafana + Evidently AI
  • Orchestrate ML pipelines with Apache Airflow
  • ML CI/CD with GitHub Actions

Weekly breakdown

WeekTopicTime
Week 1MLOps intro + MLflow + DVC10-12 hours
Week 2FastAPI serving + Docker + K8s12-15 hours
Week 3Monitoring + CI/CD10-12 hours
Week 4Airflow + End-to-End capstone12-15 hours

Chapter order

  1. Introduction to MLOps
  2. MLflow — Experiment tracking
  3. DVC — Data Versioning
  4. FastAPI + ML Serving
  5. Docker and Kubernetes
  6. Model Monitoring
  7. CI/CD for ML
  8. Airflow and Prefect
  9. Exercises

What can you do by the end of the month?

  • Build a full production ML system: training → versioning → serving → monitoring
  • ML model deployment on Kubernetes
  • Automatically detect model degradation via drift detection
  • CI/CD pipeline for ML (test, validate, deploy)
  • Weekly retraining via Airflow DAG
  • Meet the MLOps Engineer requirements written in job descriptions

Tip for Backend Dev — this month is your golden month!

It’s in this month that your existing knowledge gives you a strong advantage:

Backend knowledgeApplication in MLOps
Docker, docker-composeML containers
PostgreSQLFeature store, prediction logs
RedisModel cache, feature cache
Celery, KafkaAsync inference, streaming
GitHub Actions / GitLab CIML CI/CD
Nginx, load balancingML model serving
Prometheus, GrafanaML monitoring
REST API designML inference endpoints
Async/awaitConcurrent inference
MicroservicesML services architecture

Most ML Engineers (coming from data scientist roles) learn these things from zero. Your starting position is much higher than theirs.

Cloud Cost (optional)

This month you’ll need cloud services. Options:

  1. AWS Free Tier($300 credit for new accounts)
  2. GCP Free Tier($300 credit)
  3. DigitalOcean($200 credit student/coupon)
  4. Hetzner — cheapest (€5/month server)
  5. Local Kubernetes(minikube, kind, k3s) — free, enough for small projects

**Advice:**main exercises on local Docker + minikube, use real cloud only for capstone.

Start

Start with Introduction to MLOps.