AI Handbooks
ai.zenoagi.com
A field guide to modern AI · 9 volumes · continuously revised
Free, systematic, engineering-first knowledge bases — from the mathematics of machine learning to the skeleton of an AI agent. Pick a volume, or follow a route below.
9 / 9 volumes
The foundations — learning from data.
Supervised & unsupervised learning, model evaluation, feature engineering, and MLOps. The base layer everything else on this site is built on.
Neural networks, in depth.
Backpropagation, optimization & regularization, attention mechanisms, CNNs & Transformers, generative models, and the road to LLMs.
How large language models work, end to end.
Language modeling, the Transformer, pretraining, scaling laws, fine-tuning & alignment, evaluation, prompting, RAG, agents, and deployment.
Learning by trial, error, and reward.
MDPs, value learning, policy gradients, Actor-Critic, offline RL, multi-agent systems, RLHF, and world models — the algorithms behind alignment.
Make models fast — from kernels to serving.
Performance & bottlenecks, quantization & compression, kernels & computation graphs, serving & orchestration, hardware, and engine case studies.
From checkpoint to production.
Inference engines, model formats & compression, serving & deployment architecture, performance optimization, monitoring, and MLOps practice.
Principles → practice of autonomous agents.
The agent loop, prompt & context engineering, tools & MCP, memory, planning, multi-agent systems, evaluation, security, and framework selection.
Anatomy of agent skeletons.
How agent harnesses are designed and built: the loop, context engineering, tools, memory, subagents, safety, and evaluation — from reading code to writing your own.
A map of what’s hot, and why.
Concept maps and the underlying tech behind today’s hottest fields: LLMs, Transformers, prompting, RAG, agents, multimodal, diffusion, alignment, and safety.
Nothing in the catalog matches “”.
Not sure where to start? Four curated paths through the catalog.
Zero to AI engineer
Build the foundation first, then stack deep learning and language models on top of it.
Ship models to production
Understand the model, make it fast, then put it behind a reliable serving stack.
Build & understand agents
Learn what makes agents work, then take apart the harnesses that power the best coding agents.
Catch up with the field
A fast survey of the hot concepts, with paper deep-dives to go further where it hooks you.
AI Handbooks is a set of nine independent, systematically organized knowledge bases: methodology first, then core concepts, case studies, paper walkthroughs, hands-on practice, career paths, and glossaries. Every volume is written to be read in order — or consulted as a reference.
Each handbook lives at its own path on this domain — /ml/, /llm/,
/agent/ and so on — built as a static site you can read, search, and bookmark
freely. Start from the catalog above, or jump straight into a volume.