Foundation bridge

40 hours. Only if the entry diagnostic put you here. It compresses the code and math on-ramp (A1 and A2) into five blocks with an exit gate; classical ML then comes in Module 02. If a block feels slow, switch to the full Path A module it names.

Who needs this

People who can code a little but have not used Git, tests, HTTP APIs or SQL daily, or whose math stopped before gradients. Not for people who have never programmed; they should take Path A from A1.

Block 1: Python and the command line, 10 hours

Typing, functions, classes, modules, virtual environments, exceptions, reading tracebacks. Navigate, pipe and grep in a terminal. Resource: Harvard CS50P weeks 0 to 4 or the official tutorial, plus 15 Exercism problems with mentor feedback. Full version: Path A, module A1.

Block 2: Git and testing, 8 hours

Clone, branch, commit, push, pull request, resolve one merge conflict on purpose. pytest: write tests first for one small function, run them, break them. Resource: MIT Missing Semester lectures 1 to 6.

Block 3: HTTP, JSON and Docker, 8 hours

What a request is, status codes, JSON, a FastAPI endpoint with two routes, a Dockerfile, run it locally and hit it with curl. Resource: FastAPI tutorial, Docker Getting Started.

Block 4: SQL and just-enough math, 10 hours

SELECT, JOIN, GROUP BY, window functions on a CSV loaded into SQLite. Then 3Blue1Brown Essence of Linear Algebra episodes 1 to 4 and Essence of Calculus episodes 1 to 3, and StatQuest on bias/variance and confusion matrices. Run labs/lab01_gradient_descent.py and rewrite it from memory. Full version: Path A, module A2.

Block 5: exit gate, 4 hours

Retake the entry diagnostic. Pass bar is 8 of 10 questions and all four tasks. Then start Module 01.