datatalksclub/mlops-zoomcamp

★ 14,606Jupyter NotebookAudience · dataComplexity · 4/5Setup · hard

Mindmap

mindmap
    root((mlops-zoomcamp))
      Inputs
        Video lessons
        NY Taxi dataset
        Code notebooks
      Outputs
        Trained models
        Deployed services
        Monitoring dashboards
      Use Cases
        Self-paced learning
        Career switch to MLOps
        Team training
      Tech Stack
        Python
        MLflow
        Docker
        Terraform
        AWS

mindmap root((mlops-zoomcamp)) Inputs Video lessons NY Taxi dataset Code notebooks Outputs Trained models Deployed services Monitoring dashboards Use Cases Self-paced learning Career switch to MLOps Team training Tech Stack Python MLflow Docker Terraform AWS

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Learn end-to-end MLOps by working through six modules and a final project at your own pace.

USE CASE 2

Deploy a taxi-fare prediction model as a Flask web service or AWS Kinesis and Lambda stream.

USE CASE 3

Set up MLflow experiment tracking and a model registry for your own ML projects.

USE CASE 4

Add Prometheus, Evidently and Grafana monitoring to an ML service to catch data drift.

Tech stack

PythonMLflowDockerFlaskAWSPrometheusGrafanaTerraform

Getting it running

Difficulty · hard Time to first run · 1day+

Course assumes Python, Docker, command line and at least a year of programming, plus AWS account for some modules.

In plain English

This repository hosts MLOps Zoomcamp, a free nine-week course from DataTalks.Club that teaches how to take a machine learning model from a notebook to a running service. MLOps stands for machine learning operations, the practice of training, deploying, and watching over ML models in production. The course covers training and experimentation, deployment, and monitoring. The README notes that the team does not plan to run the course in 2026 but the materials are still available for self-paced learning. To work through it on your own, you watch the course videos, join the DataTalks.Club Slack workspace, and consult the linked FAQ. There is an Airtable form to register interest if the course is ever offered again. The course expects some background. Listed prerequisites are Python, Docker, basic command line, prior machine learning exposure such as the team's ML Zoomcamp, and at least one year of programming experience. The structure is six modules followed by a final project. Module 1 introduces MLOps, the MLOps maturity model, and the NY Taxi dataset that runs through the lessons. Module 2 covers experiment tracking and model management with MLflow, including saving and loading models and using a model registry. Module 3 looks at workflow orchestration and ML pipelines. Module 4 deals with model deployment, contrasting online deployment (web services and streaming) with offline batch scoring, with hands-on work using Flask for a web service and AWS Kinesis plus Lambda for streaming. Module 5 covers monitoring of ML services, using Prometheus, Evidently, and Grafana for web services and Prefect, MongoDB, and Evidently for batch jobs. Module 6 walks through software engineering practices such as unit and integration tests, linting and formatting with pre-commit hooks, CI and CD with GitHub Actions, and infrastructure as code with Terraform. The final project ties the modules together in one end-to-end build. Support is provided through the #course-mlops-zoomcamp channel on the DataTalks.Club Slack, with links to question-asking guidelines and community rules. The instructors are Cristian Martinez, Alexey Grigorev, and Emeli Dral. The README also describes DataTalks.Club itself, a global online community of data practitioners that runs events, free courses, and a newsletter, with most day-to-day discussion happening in Slack.

Copy-paste prompts

Prompt 1

Walk me through the Module 1 MLOps maturity model with the NY Taxi dataset, and explain which level a typical Kaggle notebook sits at.

Prompt 2

Set up MLflow tracking locally with a SQLite backend and log a scikit-learn model from a Jupyter notebook for the taxi dataset.

Prompt 3

Convert this notebook training script into a Flask web service that loads the model from the MLflow registry: <paste script here>.

Prompt 4

Build a Prometheus and Grafana monitoring stack with Evidently for an online ML service, using docker-compose for local dev.

Prompt 5

Write a GitHub Actions CI workflow that runs pytest, ruff and pre-commit on every push for an MLOps project.

Open on GitHub → Explain another repo

← datatalksclub on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.