Recently in the notebook
Return to a revised explanation, a practical guide or an experiment. Essays keep their original publication dates in Writing.
Browse essaysRevised notes
Dates describe note revisions, not independent verification of every claim. This is a list of current articles, not a change-by-change log.
Agent Memory and Retrieval
Choose what an AI assistant should remember and retrieve, without confusing stored notes, search results, and authority.
Tool Contracts for AI Agents
Design tools whose inputs, authority, side effects, failures, and evidence remain legible to models and operators.
Data Splits and Leakage
Block target, group, temporal, and preprocessing leakage by defining prediction time, entities, and train-only pipelines.
Model Evaluation Under Distribution Shift
Start from deployment distribution, splits, metrics, thresholds, and uncertainty instead of treating one test score as universal ability.
Data Science Working Vocabulary
A compact vocabulary for discussing datasets, modeling choices, evaluation, and reproducibility without product-specific jargon.
Activations and Gated Feed-Forward Networks
Distinguish ReLU, GELU, SiLU/Swish, and GLU variants, then select nonlinearities with controlled modern-network experiments.
Attention Variants and KV-Cache Compression
How MHA, GQA, MLA, cache quantization, and token-selection methods trade memory for implementation and quality risk.
Attention Mechanism
Understand content addressing, Q/K/V, masks, multi-head variants, efficient implementations, and interpretation limits through a numerical example.
Convolutional Neural Networks
Convolutional inductive bias, channels, padding, stride, pooling, and the distinction between translation equivariance and invariance.
Linear Regression
The linear prediction model, squared-error objective, solution methods, and the assumptions that determine what its coefficients mean.
Multilayer Perceptron
Understand MLP capacity, backpropagation, optimization failures, and inductive bias through tensor shapes and a worked XOR construction.
Recurrent Neural Networks
Understand recurrent state compression, backpropagation through time, LSTM gating, and the boundary with modern state-space sequence models.