What it does

This repo is a hand-maintained curriculum of courses, books, videos, and tutorials for learning machine learning from scratch. It covers general ML, deep learning, reinforcement learning, CNNs, RNNs, and MLOps, with each section offering a short conceptual intro followed by ranked links. The author updates it to “get rid of outdated content and deprecated tools” — a rare claim for a link aggregator.

The interesting bit

The curation has a point of view. Fast.ai is labeled “opinionated,” Andrew Ng’s Coursera specialization is recommended for those who “want to know the details under the hood,” and one job-focused course is praised for arguing that knowing decision trees is “already good enough” for tabular data work. The author also admits to only knowing about half the architectures in the “Neural Network Zoo” — a small honesty that builds trust.

Key highlights

• Covers the full stack from math foundations (Coursera’s math specialization) to production (Full Stack Deep Learning, Weights & Biases MLOps course)
• Includes newer material like Hugging Face’s diffusion models course and Lightning.AI fundamentals
• Book recommendations span from the hundred-page survey to the thousand-page Goodfellow/Bengio/Courville deep learning tome
• Explicitly maintained: author states they prune dead or deprecated resources
• Free and paid resources clearly mixed, with notes on time investment and target audience

Caveats

• No clear progression path or difficulty ratings between sections — you’re on your own to sequence the material
• Some links are to commercial platforms (Coursera, Udacity) with paywalls not always flagged
• The “update regularly” claim is hard to verify; no changelog or last-updated date visible in the README

Verdict

Worth bookmarking if you’re self-teaching ML and tired of blogspam roundups. Skip it if you need a structured course with assignments and deadlines — this is a bibliography, not a bootcamp.