DSWoK — Data Science Well of Knowledge

An open-source ML knowledge base: algorithms, architectures, metrics, system design, and interview preparation.

Read online at dswok.com

About

DSWoK is an Obsidian vault of interconnected notes on machine learning and data science. It's designed as a reference for practitioners to look up during work or interview preparation.

What's inside

• General ML — classical algorithms, validation, regularization, bias-variance trade-off
• Deep learning — attention, recommendation architectures, parameter-efficient training (LoRA)
• NLP — word embeddings, Transformer, BERT, Retrieval-Augmented Generation
• Metrics and losses — classification, regression, ranking/recsys, computer vision, and NLP metrics; general, NLP, and computer vision loss functions
• Interview preparation — ML fundamentals, deep learning, statistics, ML system design, behavioral, Leetcode templates
• Use cases — end-to-end design walkthroughs for production ML problems

Additionally, I write paper reviews on my blog they're intentionally not duplicated here.

How to use it

Read online. dswok.com is the easiest way — full navigation, search, and graph view.

Use locally in Obsidian. Clone the repo and open it as a vault:

git clone https://github.com/Erlemar/dswok.git

You get the full Obsidian experience: graph view, backlinks, and interactive search.

Read on GitHub. Every note is a plain markdown file. Browse the folders directly if you don't need the cross-linked view.

License

© Andrey Lukyanenko. The content is licensed under CC BY 4.0: you're free to share and adapt it, including commercially, as long as you give appropriate credit. See LICENSE for the full text.