The on-ramp: 36 weeks if you need it
This course starts at Week 1 assuming working programming comfort. If you are coming in without that base, use this 36-week prerequisite lane: nine resources, in order, with a weekly budget. Take it if you need it; skip it if you don't.
Nothing here is a hidden prerequisite. The 108-week curriculum is self-contained. But if you are missing fundamentals, this is the sequence and the time budget to close the gap before Week 1 - so the arc starts on solid ground instead of shaky ground.
How the nine resources fit into 36 weeks
| Weeks | Phase | Resources | What you can do after |
|---|---|---|---|
| W1–2 | Learning how to learn | Google Like a Pro · How To Read Documentation · Git for Beginners | Search and docs fluency plus version control - the self-serve skills every later week compounds. |
| W3–6 | Computer science foundations | CS50x Lectures 1–6 | C, memory, and algorithms - the vocabulary of computation the course assumes at Week 1. |
| W7–10 | A computer from first principles | Nand to Tetris | From NAND gates to a CPU, compiler, and OS - nothing is magic anymore. |
| W11–16 | Mathematics for computer science | MIT 6.042J | Proofs, induction, counting, probability, and graphs - the exact reasoning Weeks 7–30 rely on. |
| W17–18 | Programming paradigms | YACR | Procedural, functional, OO, and let-it-crash - the lenses behind every framework you will read. |
| W19–28 | Statistical learning | Statistical Learning with Python | Supervised and unsupervised learning - the empirical core of the ML and evaluation weeks. |
| W29–32 | Software engineering at scale | Software Engineering at Google | Design docs, testing, code review, and ownership - the discipline that makes weekly artifacts professional. |
| W33–36 | Synthesis project | All of the above | Consolidate everything into one working project - then Week 1 of the arc starts on solid ground. |
Nine links, in the order to take them
Self-serve research: filter, scope, and verify results instead of hoping Google guesses right.
Navigate any framework docs - quickstarts, references, and changelogs - without a tutorial.
Commit, branch, merge, and recover - the tooling habit every weekly artifact depends on.
C, memory, algorithms, data structures, and web - the shared vocabulary of the whole course.
Build a working computer from NAND gates up - assembler, VM, compiler, OS - and a Tetris game.
Proofs, induction, counting, probability, and graphs - the reasoning the course front-loads in Weeks 7–30.
Procedural, functional, object-oriented, and let-it-crash thinking - and why each exists.
Supervised and unsupervised learning from the data side - the empirical backbone of every ML week.
How teams design, test, review, and ship at scale - the professional layer under the technical arc.
The 36-week lane puts mathematics first and engineering practice second, which matches the order of the main course. When it's done, you enter the tensor-to-tenant arc - see the curriculum for what Week 1 looks like.