ML Interview Notes
Build status

Build status

What is live on the site today, and what is still to be written. This page is generated from content/roadmap.yml — update that file and it updates here.

10 of 21 tracked items complete

Platform

The site itself — how pages are built, verified, and shipped.

8 of 10 done
  • Static site generatorDone

    Markdown in content/ builds to HTML in _site/. Multi-course aware: adding a course needs a directory, a course.yml, and Markdown files — no code change.

  • Design systemDone

    Blueprint theme with per-track accent colours, DM Sans / Fira Code, and full dark mode driven by the viewer's system setting.

  • Math, diagrams, and code renderingDone

    KaTeX for inline and display math; Mermaid diagrams with pan, zoom, and full-size view; syntax-labelled code blocks; scrollable tables.

  • Continuous deploymentDone

    Pushes to main build and publish to GitHub Pages automatically. Pull requests run the same build without deploying.

  • Build verificationDone

    Every deploy is gated on a check suite asserting that each source file produced a page, every internal link resolves, every diagram rendered, and math survived the pipeline. A static site fails silently; this is what stops a broken link from shipping.

  • Custom domain and HTTPSDone

    mlinterviewnotes.com, served over TLS.

  • Diagram correctness auditDone

    All 67 course diagrams were checked twice — mechanically, that every one parses and renders; and semantically, that each agrees with the prose around it and with the reference implementation. 55 were correct as written, 12 carried inaccuracies, and all 12 are now fixed. No diagram was found to teach anything false.

  • Interactive widget supportDone

    Pages can embed raw HTML blocks and pull in their own scripts, so D3-driven explanations live alongside ordinary Markdown.

  • Site-wide searchPlanned

    Client-side index across all courses and notes. No backend needed at this scale.

  • Contribution guidePlanned

    A CONTRIBUTING.md covering the Markdown conventions, how to preview locally, and what the verification pass checks.

Courses

Long-form, sequential material built from first principles.

1 of 3 done
  • Transformers Deep Dive — 17 modulesDone

    From "why did we abandon RNNs?" through to the configuration choices in production 2026 models. Includes 67 diagrams, worked numeric examples, self-check questions, and a runnable reference decoder.

  • Classical ML refresherPlanned

    Linear and logistic regression, trees and ensembles, SVMs, clustering — derived rather than recited.

  • LLM training and fine-tuningPlanned

    Pretraining, SFT, preference optimisation, LoRA and friends.

Notes

Topic-by-topic reference notes. Structure is in place; the writing is not. Each topic below is a page waiting to be written.

1 of 6 done
  • Math for ML — 1 of 5 writtenIn progress

    Linear Algebra is written, with five interactive visualizations — recovered from the previous site, where it was the only topic that had real content. Calculus, probability and statistics, optimization, and discrete mathematics are still to come.

  • Libraries for ML — 6 topicsIn progress

    Structure listed; content pending.

  • Machine Learning — 6 topicsIn progress

    Structure listed; content pending.

  • Deep Learning — 6 topicsIn progress

    Structure listed; content pending.

  • NLP — 6 topicsIn progress

    Structure listed; content pending.

  • One page per topicDone

    Notes are two levels: a category page lists its topics, and each written topic is its own page beneath it — /notes/math/ holds /notes/math/linear-algebra/. A topic never sits beside the category it belongs to.

Sections still to build

Planned in the original site and carried forward here. Both were empty in the previous database, so both start from scratch.

0 of 2 done
  • Question & AnswerPlanned

    Interview questions with worked answers, explanations, and code — organised by category and difficulty.

  • System DesignPlanned

    ML system design patterns: problem statement, architecture, key components, scaling considerations, and trade-offs.