Week 3 · Season 1 · Spark
Tools of the trade
You can work in the terminal, organise code into functions and modules, read errors calmly, use Git and GitHub, use AI tools well, and plan a feature before you build it.
This week
- The terminal, folders and virtual environments
- Functions, modules and reading errors
- Git commits, branches and GitHub
- Using AI tools well, and using AI to learn
- Requirements, boxes and arrows, the one-page plan
How you learn
Break something on purpose and fix it with Git. Ask an AI tool for code, then find the bug it wrote before you run it.
What you build
FRIDAY gets a home: a GitHub repo with a README, a plan, a diagram and a clean history of small commits.
Stretch: Add a GitHub Action that runs your tests on every push.
What you ship
- A public repo a stranger can run from the README
- A plan and diagram for FRIDAY's next part
Lessons
- The keyboard cockpit
After this deck you can open a terminal (a window where you type commands), find your way around folders with
pwd,ls,cdandmkdir, run a script from its own folder withpython3 friday.py, and give a project its own virtual environment for the packages it installs. - Recipes you can reuse
After this deck you can write a function (a named recipe) with
defandreturn, move functions into a module youimport, and read a traceback (Python's error note) from the bottom up to fix the three errors beginners meet first. - A time machine for code
After this deck you can save your work as commits (snapshots with a note), throw away a mess with
git restore, undo a bad commit withgit revert, try risky ideas on a branch, and push FRIDAY's history to a repo on GitHub. - The brilliant intern who never says "I don't know"
After this deck you can brief an AI coding tool with context (what, where, what done looks like, what not to touch), ask it to plan before it edits, verify what it wrote by reading and running every line, and use it to learn by making it quiz you instead of handing you answers.
- Draw it before you build it
After this deck you can turn a wish into requirements (sentences that say what done looks like), draw your project as a diagram of boxes and arrows, and fit a feature into a one-page plan with a time budget, a must-have list and a not-now list.
- Go deeper: the intern gets the keys
After this deck you can explain how an agentic coding tool works in a loop, write a short project memory file, and review an AI's pull request line by line before you merge.
Labs
Python in the browser, next to the lessons.
Weekend build
- FRIDAY, part 2: a home and a personality
Give FRIDAY a proper home on GitHub, then a personality, rules and answers you can trust.
Skills you prove
- Terminal & Git
- Work in the terminal and keep your code safe with Git commits
Free links for this week
Hand-picked official docs, papers and talks.
- 3Blue1Brown: Neural networks & transformers · course
- Karpathy: Neural Networks: Zero to Hero · course
- micrograd · repo
- minbpe · repo
- Attention Is All You Need · paper
- Keshav: How to read a paper · paper
- PyTorch tutorials · docs
- Raschka: LLMs from scratch (code) · repo