Machine Learning

Machine Learning Specialization

Andrew Ng's updated three-course ML specialization — covering supervised learning, unsupervised learning, and reinforcement learning with modern Python tools. The most widely taken ML course in the world; the benchmark for team-wide ML onboarding.

Practical Deep Learning for Coders

Fast.ai's top-down approach to deep learning — start building real models immediately and learn theory as needed. Used by engineers transitioning into ML roles who benefit from hands-on practice over lectures. Free and actively maintained.

CS229: Machine Learning

Stanford's graduate-level ML course covering the mathematical foundations of supervised learning, Bayesian methods, SVMs, and unsupervised learning. The rigorous technical baseline for engineers who need to understand model behavior, not just call APIs.

An Introduction to Statistical Learning

The gold standard introduction to statistical learning — free in PDF, with Python and R labs. Covers linear regression, classification, resampling, regularization, and tree methods with mathematical rigor and practical examples. Required reading for ML practitioners.

Designing Machine Learning Systems

Chip Huyen's systems-level treatment of production ML — data management, feature engineering, model selection, deployment, and monitoring. Addresses reliability, scalability, and adaptability requirements for ML systems in regulated enterprise environments.

Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow, 3rd Edition

The most widely used practical ML book — covering the full stack from classical algorithms through deep neural networks. Updated for TensorFlow 2 and Keras. The benchmark reference for engineers implementing ML models in enterprise production environments.

But what is a neural network? (Deep Learning Chapter 1)

3Blue1Brown's iconic first video on neural networks — building intuition for how networks represent functions and learn from data with visual clarity unmatched in technical education. The starting point for non-engineers learning to think about ML.

Stanford CS229: Machine Learning (Full Course, Autumn 2018)

Andrew Ng's complete Stanford ML course — covering supervised learning, generative algorithms, SVMs, neural networks, and unsupervised learning with full mathematical derivations. The authoritative academic ML course available free on YouTube.

Neural Networks: Zero to Hero

Karpathy's complete series building neural networks from scratch in Python — from micrograd through character-level language models to a full GPT. The gold standard for technical leaders who want deep understanding of how modern ML systems work.

Improving RCT-Based Treatment Effect Estimation Under Covariate Mismatch via Calibrated Alignment

ChatGPT is now a partner for your most ambitious work

MultiFair: Multimodal Balanced Fairness-Aware Medical Classification with Dual-Level Gradient Modulation

Beware What You Autocomplete: Forensic Attribution of Backdoored Code Completions

AutoAnchor: Stable Diffusion Unlearning Using Cross-Attention as a Manifold Surrogate

Who Analyses the Analyser? Self-Validating LLM Hazard Analysis with Constitutional Meta-STPA

Who Gets Missed in the Tail? Thresholded Subgroup Underdiagnosis in Long-Tailed Chest X-ray Classification

Our approach to government and national security partnerships

Vision Foundation Models in Radiology: A Scoping Review of Data, Methodology, Evaluation and Clinical Translation

Multi-Agent AI Control: Distributed Attacks Hamper Per-Instance Monitors

Safe Reinforcement Learning using Ideas from Model Predictive Control

Predicting LLM Safety Before Release by Simulating Deployment

Kaggle

Google's data science platform providing free GPU compute, public datasets, competitions, and a library of community notebooks. The standard learning environment for ML practitioners who learn by doing — and a useful benchmark for team ML skill levels.

PyTorch

Facebook's open-source deep learning framework — now the dominant choice for both research and production ML. Dynamic computation graphs, strong ecosystem, and Hugging Face integration make it the default framework for modern neural network development.

scikit-learn

The most widely-used Python ML library — providing simple, consistent APIs for classification, regression, clustering, and model selection. The standard tool for ML practitioners implementing classical algorithms with production-grade reliability and documentation.