Skip to content
Cart

Methodology

The same format in every module, end to end. Two to four hours per module in total.

The five steps of every module

  1. Lean theory

    The concept, the visual intuition, the why.

    10–20 min of video

  2. Code-along

    You build alongside the student, making mistakes on purpose and fixing them.

    30–60 min

  3. Hands-on challenge

    Same problem, different dataset and context. No copy-paste.

    1–2h of student work

  4. Walkthrough of the solution

    The official solution plus the most common mistakes.

    10–15 min

  5. Knowledge check

    A short quiz plus submission of the code on GitHub.

    At the end of the module

The rules that back the promise

  • No level ends without an integrative project built from scratch, using everything covered.
  • Every project ends published — on GitHub, and deployed where that makes sense. Students leave with a portfolio, not just a certificate.
  • Roughly 30% theory and 70% hands-on work in every module.
  • Mathematical theory is always taught as visual intuition on video, with an optional deeper PDF for those who want it.

Why the programme is international by design

Internationalisation is an architectural decision in the curriculum, not a translation bolted on afterwards.

  • Code, variable names and comments in English from the very first module. Only the narration changes language.
  • No text burned into the video: slides and overlays stay editable separately, so translating never requires re-recording.
  • Neutral international datasets wherever possible, with local variations in the challenges.
  • Language order: Portuguese, then English, then Spanish.
  • Subtitles in every language, always.

Three languages, one recording

The programme is international by a production decision, not by translation added afterwards. The rules below apply from the very first module recorded, and they are precisely what makes Portuguese, English and Spanish viable from a single recording.

  1. AI dubbing with voice cloning

    AI dubbing tools clone the instructor's voice and sync the lips. That is why every recording is made with clean audio and no background music: it is that constraint at capture time which makes the dubbing possible later.

  2. No text burned into the video

    Slides and overlays live in files separate from the raw video. Translating the slides never requires re-recording the lesson.

  3. Code always in English

    Code, variable names and comments in English from the first module onwards. Only the narration changes language.

  4. Localisable datasets

    Neutral international datasets wherever possible; in the challenges, local variations with a translated brief.

  5. Language order: PT → EN → ES

    Portuguese, English and Spanish cover most of the initial market. Further languages come in on demand, not in anticipation.

  6. Subtitles in every language

    Beyond dubbing, subtitles in every language: near-zero cost, and what gives accessibility and searchability inside the video.

This site is already published in all three languages. No lesson has been recorded yet — in any language. The rules above describe the standard the first recording will have to meet; once a module is recorded, its status in the curriculum will state which languages it exists in.