An open-source degree plan for the self-taught AI engineer

What it does This repo is essentially a syllabus architecture project. It collects freely available courses, books, and materials into a modular, self-paced curriculum meant to take a learner from first code to building autonomous AI agents. Think of it as the course catalog and schedule for a degree that has no campus, no tuition, and no professors on payroll.

The interesting bit The hard part of self-teaching isn’t finding resources—it is knowing what to study in what order and for how long. This project solves the sequencing problem with estimated durations (six to eight weeks for math, one to two for RAG) and a capstone phase meant to yield portfolio pieces rather than just completion certificates.

Key highlights

• Seven-stage progression: Programming foundations → Math → ML → Deep Learning → Specializations (NLP, CV, RL) → Modern AI (LLMs, Agents, RAG) → Capstone projects
• Time-boxed modules with explicit week estimates, so you can budget attention like a semester planner
• Explicitly built around external heavyweights: MIT OCW, Stanford CS231n, Fast.ai, DeepLearning.AI, and Hugging Face
• Community-curated and MIT licensed, with contribution guidelines for keeping the path current
• Targets “portfolio-ready” outcomes, not just theory accumulation

Caveats

• The promised visual roadmap is listed as “coming soon,” so the current experience is entirely markdown tables
• Every learning resource lives outside the repo; this is pure curriculum architecture, not hosted coursework
• The time estimates assume significant weekly commitment—self-paced can easily mean half-pace

Verdict Worth bookmarking if you are a disciplined self-learner trying to bootstrap an AI engineering career without a bootcamp invoice. If you already have a CS foundation or need graded accountability, this will feel like a very detailed reading list.