Why this roadmap starts with Linux

Spark, GPUs, and notebooks all sit on a prompt. We teach that prompt first, and we never run it for you.

Most Data Engineering and AI curricula start at a library: pandas, PySpark, PyTorch. Those libraries are the job. They are not the ground.

The ground is a Linux machine. Extract jobs write to directories. Training runs fill disks with checkpoints. Airflow fails because a user cannot write /data. GPU boxes are reached with SSH. When something breaks at 2 a.m., you are in a terminal, reading a log, not in a slide deck.

Beginners often get stuck before Spark: they cannot find the file, cannot tell stdout from a hang, cannot explain Permission denied. We would rather spend a few hours there than pretend a notebook on Windows is the production environment.

So DCubes starts with Linux fundamentals. Absolute beginners. Windows through WSL, macOS and Linux through a real terminal.

Guidance only

This site does not run your commands. There is no in-browser terminal and no sandbox on our host. You type locally. That is inconvenient for five minutes and honest for the career: the skill is operating a machine that is yours (or yours via SSH).

If a lesson cannot be practiced that way, we will wait until we can write it that way.

What “done” with Linux means

Not LPIC. Not a sysadmin course. After the module you can:

set up a prompt
navigate a filesystem
inspect logs and CSVs
compose pipes
notice disk and memory
generate an SSH key without giving away the private half

Then Docker and Compose, then Python, SQL, and Git. Data Engineering comes after that shared spine. Linux is node one because everything else logs in through it.