Skip to content

Don't learn tech. Build it.

Nine months from your first commit to a portfolio you can put in front of an employer. You'll learn to work with data, train models, and build with AI — in that order, because each one rests on the last.

Start here

What you'll become

Three roles the industry hires for, plus the habits that get you hired. You work through the roles in sequence, and each one leaves you with skills that stand on their own.

  • Data Analyst

    Get data out of messy sources and make it say something useful. Python, pandas, SQL, visualisation, statistics, and the classic models that still solve most business problems.

  • Machine Learning Engineer

    Go deeper than the classic models. Neural networks, computer vision, and the work of training something well and getting it running in production.

  • AI Engineer

    Build products on top of language models. NLP, retrieval augmented generation, agents, and the tooling that connects models to real systems.

  • Ready for the room

    Explain your work to people who do not code. Handle a technical interview and a whiteboard. Pick up an unfamiliar tool quickly, because the one you learn today will be replaced.

The path

The nine-month path Three stages over nine months. First, Data Analyst: Python, pandas, SQL, visualisation, statistics and classic machine learning models. Second, Machine Learning Engineer: deep learning, computer vision, training and deployment. Third, AI Engineer: natural language processing, retrieval augmented generation, agents and shipping products built on models. The programme ends with a portfolio project. 1 Data Analyst Months 1–3 Python, pandas, SQL, visualisation, statistics, classic ML models 2 Machine Learning Engineer Months 4–6 Deep learning, computer vision, training at scale, deployment 3 AI Engineer Months 7–9 NLP, RAG, agents, MCP, shipping products built on models A portfolio project, and the interview that follows

Each stage assumes the one above it. Skipping ahead usually costs more time than it saves.

See the full roadmap

How to work through it

Read less, build more

Every tutorial ends with something running on your machine. If you finish a page and nothing runs, you skipped the part that mattered.

  • Type the code out rather than copying it. The mistakes you make are the lesson.
  • Keep a short note after each page: what you built, what broke, how you fixed it.
  • When an example works, change something and predict the result before you run it.
  • Explain a concept to a classmate within a day of learning it. Teaching is the test.

Stuck?

Ask in the cohort channel before you lose an hour to it. If you spot an error on a page, use the pencil icon at the top right to open a pull request — the handbook improves because trainees fix it.

Built for Holberton School Albania. Questions and corrections: evis.plaku@holbertonschool.com or open an issue.