A pirate library for the self-taught ML engineer

What it does

lit is a curated GitHub repository of technical books for self-taught AI practitioners, organized by topic from AI fundamentals to math, CS, and systems. Each entry includes a personal review and often a direct PDF link hosted in the repo itself.

The interesting bit

The curation is aggressively opinionated and occasionally petty. The maintainer admits some books may not be “acquired in a lawful way,” and includes one author’s work specifically because that author plagiarized his machine learning app. It’s a reading list with a vendetta.

Key highlights

• Covers the canonical texts: Speech and Language Processing, Deep Learning Book, AIMA, Introduction to Algorithms, SICP
• Includes practical starters like Hands-On Machine Learning and Learn Python the Hard Way
• Personal, informal reviews that actually say which books are skimmable vs. essential
• A wiki page for total beginners who don’t know where to start
• Topics span NLP, data science, programming, OS/Unix, and math/stats

Caveats

• Many books are hosted as PDFs directly in the repository, with explicit acknowledgment that legality is uncertain
• The maintainer’s background is NLP/speech-heavy, so the AI section skews accordingly
• Some entries are incomplete (the README truncates mid-sentence on The Princeton Companion to Mathematics)

Verdict

Worth browsing if you’re building a self-directed CS/ML curriculum and want honest takes on which thousand-page textbooks to actually finish. Skip it if you need rigorous legal copies or a balanced view of fields outside NLP.