AI

Topics

• LLM model comparison: 379 releases and size variants with creators, dates, parameters, licenses, local memory estimates, and model links

• Prompt Engineering: crafting LLM inputs to reliably elicit a target output

• Natural Language Processing: classical NLP, spaCy pipelines, NLTK corpus analysis, sparse classifiers, and where LLMs fit

• AI Harness Development: the scaffolding around an LLM that turns it into something useful

• LangGraph: stateful graph workflows and a Python MCP server example

• RAG (Retrieval-Augmented Generation): grounding LLM responses in external knowledge

• AI Skill Development: packaged, invocable capabilities loaded on demand

• AI Coding Tool Blindspots: predictable failure modes of modern AI coding assistants

• DESIGN.md: Google Labs’ open-source file format for teaching coding agents your visual identity

• Context Development Lifecycle: Patrick Debois’s framework for treating organizational context as a managed artifact with its own generate-evaluate-distribute-observe loop

• Model Distillation: training capable smaller models from larger ones — the theory, LLM methods, distillation prompts, case studies (Alpaca, Vicuna, Orca, Phi, DeepSeek-R1-Distill), and the economic implications

• LLM reasoning benchmarks and metrics: 8-part series on what the leaderboard numbers actually mean

• AI text tells: the patterns that reveal LLM-generated text and the training dynamics that produce each one

• LLMs vs. agentic AI vs. AI agents: three distinct concepts: the base model, the loop-plus-tools architectural pattern, and specific implementations

• Deep Learn skill: a Claude Code skill implementing 15 learning science patterns for durable retention of any knowledge base

Each topic has a hub overview, subtopics for deeper dives, and references to authoritative external resources.