MLOps

roadmap.sh: https://roadmap.sh/mlops

Suggested path through the MLOps nodes. Each node links to its lesson when written.

Nodes

Foundations

• What is MLOps
• MLOps Principles
• MLOps Components

Programming

• Programming Fundamentals
• Python
• Go
• Bash

Version Control

• Version Control Systems
• Git
• GitHub
• GitLab

Maths & ML Fundamentals

• Maths & Statistics
• Machine Learning Fundamentals
• Machine Learning
• Deep Learning

ML Frameworks & Libraries

• scikit-learn
• TensorFlow
• PyTorch

Data Engineering

• Data Engineering Fundamentals
• SQL
• Data Pipelines
• Data Ingestion Architecture
• Data Lakes & Warehouses
• Data Lineage
• Data Versioning (DVC)
• CML

Data Processing & Streaming

• Spark
• Flink
• Kafka

Experimentation & Tracking

• Experiment Tracking
• MLflow
• Model Evaluation

Containerization

• Containerization
• Docker
• Kubernetes

Cloud Computing

• Cloud Computing
• AWS / Azure / GCP
• Cloud Native ML Services

Infrastructure as Code

• Infrastructure as Code
• Terraform
• Ansible

CI/CD

• CI/CD
• GitHub Actions
• Jenkins

Orchestration & Deployment

• Orchestration
• Orchestration & Deployment
• Airflow
• Kubeflow
• Model Training & Serving

Monitoring & Observability

• Monitoring & Observability
• Prometheus
• Grafana

Explainability

• Explainable AI
• SHAP
• LIME

Edge AI

• Edge AI
• Jetson
• PyTorch Mobile
• TFLite

Resources

See resources.md.

Project ideas

• Build an end-to-end ML pipeline: ingest data with a Kafka/Spark stage, train a scikit-learn model, track runs in MLflow, and serve it behind a containerized API on Kubernetes.
• Set up a CI/CD workflow with GitHub Actions + DVC that retrains and redeploys a model on new data, gated by automated model-evaluation checks before promotion.
• Stand up a monitoring stack (Prometheus + Grafana) for a deployed model that tracks latency, request volume, and data/concept drift, with alerting that triggers retraining.