Foundations
Foundations I can reuse under pressure.
This part of the blog turns strong existing resources into my own working knowledge: concise explanations, bilingual notes, practical examples, and links back to the sources that shaped the post.
Featured series
Statistics for AI Research
A bilingual series for uncertainty, confidence intervals, bootstrap, hypothesis testing, effect size, and statistical testing for NLP/LLM evaluation.
Featured series
Machine Learning Foundations
A practical bilingual series for interviews and applied research: splits, overfitting, cross-validation, metrics, model selection, search, leakage, and representation learning.
Featured series
NLP & LLM Foundations
A research-oriented bilingual series for tokenization, embeddings, seq2seq, attention, Transformers, fine-tuning, PEFT, prompting, and LLM evaluation.
Foundation tracks
Each track is broad enough for interviews and deep enough to support research writing.
Mathematics
Vectors, matrices, embeddings, dot products, optimization, gradients, PCA, and graph ideas used in ML/NLP.
Statistics
Probability, distributions, confidence intervals, hypothesis tests, bootstrap resampling, and experimental uncertainty.
Machine Learning
Train/validation/test splits, overfitting, cross-validation, regularization, model selection, losses, and tuning.
NLP & LLMs
Tokenization, embeddings, Transformers, fine-tuning, prompting, generation, and NLP evaluation basics.
Research Engineering
How experiments become reproducible: configs, logs, evaluation tables, data checks, demos, and documentation.