When three minutes isn't enough.

Google's own ML primer with interactive exercises. The fastest credible route from "no idea" to "I can read a model card".

LLMs, embeddings, responsible AI and Vertex. Feeds directly into the Gen AI Leader exam.

Builds a transformer from scratch on video. Hard, worth it, and the best cure for treating models as magic.

If backpropagation has never clicked, watch this before reading another explainer.

Practical and code-first. Fine-tuning, tokenizers, datasets — for PMs who want to run something, not just discuss it.

The reference I keep open. Techniques, failure modes and papers, kept current — better than any paid prompt course.

Built with the model vendors. The fastest way to stay current on a specific technique.

Top-down teaching — you train a model in lesson one. Needs real Python.

Short notebook-based micro-courses. Intro to ML and Feature Engineering are the useful two.

Vendor material that is unusually candid about limitations. Good on prompting and agent design.

Introductory and applied tracks. Useful for bringing a non-technical team up to a shared vocabulary.

A structured twelve-week curriculum on GitHub. Dry, thorough, and genuinely complete.

The closest thing to a canon for empowered product teams. Start with "Product vs. Feature Teams" and work backwards.

Deep operator interviews and benchmarks. The free archive alone beats most paid courses.

Product analytics taught properly — retention, funnels, cohorts, north-star design. Tool-specific but the thinking transfers.

Real PRDs, strategy docs and launch plans from real companies. Steal the structure, not the content.

Conference talks and articles. Good for breadth and for spotting what's current.

Practical templates and market breakdowns. Strong on the craft end rather than the philosophy end.

The threat list every AI PM should be able to recite. Prompt injection, data leakage, supply chain, excessive agency.

What your governance and legal teams are going to ask about. Read the core functions; skim the rest.