AI ML Jargon (github.com)

A practitioner‑first glossary of core AI/ML jargon has been published: plain‑English, punchy definitions and quick mental models (with Rick & Morty asides) aimed at people doing the work—code reviews, design docs, and late‑night debugging. The collection eschews fluff for signal, peppering entries with tiny Python snippets, practical gotchas and “don’t overfit, Morty” style warnings so readers can apply concepts immediately rather than just memorizing vocabulary. Its value to the AI/ML community is pragmatic: it bridges academic terms and production realities, speeding onboarding and improving team communication. The glossary spans fundamentals (supervised/unsupervised learning, RL), models and components (transformers, diffusion models, GANs, autoencoders, embeddings, attention), training mechanics (losses, optimizers, backprop, batch/layer norm, regularization, hyperparameters), evaluation and metrics (accuracy/precision/recall/F1, ROC/AUC, perplexity), deployment/MLOps topics (inference, latency/throughput, model registry, feature store), and efficiency tricks (pruning, quantization, distillation). By combining concise definitions, mental models and code-level tips, it makes tradeoffs and common pitfalls (overfitting, validation leakage, exploration vs. exploitation, tuning strategies) actionable—useful for practitioners who need fast, operationally relevant clarity rather than theoretical depth.