Data Engineering Zoomcamp

What it is

Data Engineering Zoomcamp is a free course run by DataTalks.Club, published as a GitHub repository with recorded videos, notes, homework and a final project. It runs as a live cohort once a year (the repository lists the next one as starting in January 2027) and can be followed self paced at any time.

What it covers

The syllabus is a tour of a realistic modern data platform, one module per week:

• Containerisation and infrastructure as code with Docker, PostgreSQL and Terraform.
• Workflow orchestration (the current edition uses Kestra).
• Data warehousing with Google BigQuery, including partitioning and clustering.
• Analytics engineering with dbt.
• Batch processing with Apache Spark.
• Streaming with Kafka.
• Workshops on data ingestion and on newer tools such as DuckDB.

The tools change a little between years; the ideas (containers, infrastructure as code, orchestration, a warehouse, transformation, batch and streaming) do not.

Prerequisites

Basic coding and some SQL. Python helps. No previous data engineering is needed, but if you have never used a terminal or Git, spend a week on those first, because every module assumes them.

Free versus certificate

Everything is free. Certificates are only given to people who complete the final project and its peer reviews during a live cohort. Self-paced learners can do all the homework but do not receive a certificate. Honestly, the certificate matters much less than the project you build and can explain.

Cloud costs

Several modules use Google Cloud. New Google Cloud accounts come with trial credit and BigQuery has a free tier, but a card is usually required and forgotten resources can cost money. Set a budget alert on day one and delete what you are not using.

Is it worth your time?

Yes, if you commit to building. It is the closest free thing to a first data engineering job: messy public data (New York taxi trips), real tools and a final project that you design, document and have reviewed by peers. The trap is watching videos without running the code, or copying the reference pipeline for your final project instead of choosing your own dataset.

Time needed

The cohort runs about nine weeks of lessons plus two to three weeks for the project. Plan on roughly 8 to 12 hours a week; more if Docker and SQL are new.

Common questions

Do I need to use Google Cloud? The course is built around it, but many learners run parts locally or adapt them to AWS or Azure. The concepts carry across.

Is the community useful? Yes. The DataTalks.Club Slack has channels for each course, and asking specific questions there is often faster than searching.