An interconnected ML reference for practitioners: core algorithms, deep learning, NLP, metrics, system design, and interview prep.
Maintained by Andrey Lukyanenko — Machine Learning Engineer at Meta (London), Kaggle Competition Master and Notebook Grandmaster, Google Developer Expert.
New here? Start with Recommendation system — it's the most comprehensive example of how I approach an end-to-end ML problem.
Browse by topic
Core ML — classical algorithms, validation, regularization, bias-variance trade-off.
Start with: Gradient boosting
Deep Learning — attention, recommendation architectures, parameter-efficient training.
Start with: Attention
NLP — from word embeddings to Transformers, BERT, and Retrieval-Augmented Generation.
Start with: RAG or Transformer
Metrics and losses — classification, regression, ranking, recommendation, computer vision, NLP.
Start with: General losses or f1 score
Interview preparation — ML, DL, statistics, system design, behavioral, Leetcode templates.
Start with: ML System design
Use cases — deep-dive design walkthroughs for production ML problems.
Start with: Recommendation system
Looking for paper reviews?
I’ve written 190+ paper reviews on andlukyane.com.
Open-source on GitHub.