Academy learning map

Build the core. Then specialize with intent.

KnowledgeOS combines one sequenced, twelve-volume handbook with practical course previews and focused engineering specializations. Start with AI developer tools or workplace AI, use the handbook for durable foundations, then choose the specialization that matches the systems you need to build or lead.

Core phases: 4
Handbook volumes: 12
Focused courses: 27
Available study: 218 hr 50 min

118 hr 30 min in the core handbook plus 100 hr 20 min across the focused catalog. Focused courses are elective paths, not mandatory steps.

Free previews and practice

Use AI on work you can check.

Read three complete sections of each developer-tools course at no cost. A 12-month Course Pass unlocks the complete course and its workshop. The Excel and PowerPoint course remains fully free.

How the academy fits together

One academy, two connected layers.

Core curriculum

The twelve-volume handbook supplies the concepts, mathematics, engineering progression, and architecture judgment that later topics depend on.

12 volumes · 161 chapters · 118 hr 30 min

Specialization tracks

Focused courses deepen a role or production system. Choose one after reaching its recommended core phase; complete all only when your role genuinely spans them.

27 courses · 100 hr 20 min

Find your core entry point

Start with evidence, not confidence.

Skip a core phase only when you can explain its outcomes and apply them in a design or implementation review.

New to AI or changing careers

Begin at Phase 1 and keep the sequence intact.

Comfortable with ML and deep learning

Review Phase 1 outcomes, then begin at Volume 5.

Already building RAG or agent systems

Use Phases 2 and 3 to close engineering gaps before production.

Leading platforms, risk, or architecture

Audit Phases 1-3, then use Phase 4 as the decision framework.

Chapter levels

Know what each chapter asks of you.

Each chapter declares one of five levels, shown next to its title in section and volume lists. The phases mix levels: the counts below come from the chapters themselves, with a link to the first chapter at each level.

Beginner: You can do school algebra and follow a short program, plus any earlier chapter the Before you start section names. Chapters ask you to trace each mechanism by hand and learn the vocabulary later chapters build on.
Intermediate: You know the vocabulary of earlier chapters and can run Python or Java code. Chapters ask you to implement methods, run them on real data, and choose between them from measured results.
Advanced: You can implement and evaluate the standard methods of a topic. Chapters ask you to derive how those methods work, compute their results by hand, and diagnose when they fail.
Architect: You understand the components from earlier chapters; production experience helps but is not assumed. Chapters ask you to design the whole system, budget its latency and cost, and defend each decision with numbers.
Volume review: You have worked through the chapters of the volume. The review asks you to combine them in one design exercise with real numbers and a model answer, without new mechanisms.

Levels in each phase

Phase 01Build the foundations66 chapters
Beginner
17 chapters
Start with Volume 1 Chapter 1The Dream of Thinking Machines
Intermediate
30 chapters
Start with Volume 1 Chapter 11Foundation Models
Advanced
15 chapters
Start with Volume 3 Chapter 5Ensemble Methods
Architect
0 chapters
Volume review
4 chapters
Start with Volume 1 Chapter 15End of Volume Review & Interview Questions
Phase 02Engineer generative systems39 chapters
Beginner
3 chapters
Start with Volume 6 Chapter 1Prompt Basics
Intermediate
12 chapters
Start with Volume 5 Chapter 1Tokens
Advanced
13 chapters
Start with Volume 5 Chapter 3Positional Encoding
Architect
8 chapters
Start with Volume 5 Chapter 6Pretraining
Volume review
3 chapters
Start with Volume 5 Chapter 14End of Volume Review & Interview Questions
Phase 03Design agentic workflows33 chapters
Beginner
0 chapters
Intermediate
11 chapters
Start with Volume 8 Chapter 1What is an Agent?
Advanced
11 chapters
Start with Volume 8 Chapter 5LangChain
Architect
8 chapters
Start with Volume 8 Chapter 9Enterprise Agents
Volume review
3 chapters
Start with Volume 8 Chapter 12End of Volume Review & Interview Questions
Phase 04Lead in production23 chapters
Beginner
0 chapters
Intermediate
2 chapters
Start with Volume 12 Chapter 1AI Ethics
Advanced
8 chapters
Start with Volume 11 Chapter 1AI Architecture
Architect
11 chapters
Start with Volume 11 Chapter 5AI Observability
Volume review
2 chapters
Start with Volume 11 Chapter 13Enterprise AI Architecture Capstone & Design Review

Phase 01

Build the foundations

AI history, mathematics, machine learning, and deep learning mental models.

Exit evidence: Explain how modern AI systems learn, where their behavior comes from, and how core model families differ.

Phase 02

Engineer generative systems

LLMs, prompt systems, retrieval pipelines, evaluation, and model serving.

Exit evidence: Build evidence-grounded generative applications and reason about their quality, latency, cost, and failure modes.

Phase 03

Design agentic workflows

Tools, memory, planning, MCP, orchestration, and multi-agent coordination.

Exit evidence: Design bounded agents that use tools safely, recover from failure, and remain observable under production load.

Phase 04

Lead in production

Architecture, reliability, security, cost control, governance, and risk.

Exit evidence: Make architecture and governance decisions that turn capable prototypes into accountable enterprise systems.

Focused specialization

Choose the systems you want to become responsible for.

These courses extend the core curriculum without duplicating its sequence. Follow one track for role depth, or combine tracks when your production ownership crosses boundaries.

Track 01

Build production AI applications

Turn model capabilities into dependable backend, multimodal, and real-time product experiences.

Best after Core Phase 2

Exit evidence: Ship an AI application with explicit model boundaries, grounded context, multimodal or streaming behavior, evaluation, and operational controls.

Track 02

Engineer agents and developer workflows

Progress from bounded agent loops to retrieval, tools, protocols, durable execution, and repository automation.

Best after Core Phase 2; use Core Phase 3 alongside it

Exit evidence: Design an agentic workflow that limits authority, preserves evidence, recovers from failure, and can be evaluated before release.

Track 03

Own models and AI infrastructure

Go beneath application APIs into adaptation, inference serving, GPU economics, and distributed training systems.

Best after Core Phase 2

Exit evidence: Choose and defend a model adaptation, serving, or training architecture using measured quality, throughput, latency, resilience, and cost evidence.

Track 04

Lead enterprise AI platforms

Connect governed data, reliability, security, and architecture decisions into an accountable production platform.

Best after Core Phase 3

Exit evidence: Review an enterprise AI design across data authority, evaluation, security, SLOs, failure recovery, capacity, and organizational ownership.

Track 05

Build the systems foundation underneath

Reason correctly about partial failure, consistency, partitioning, coordination and saturation before any of that infrastructure is asked to carry a model.

Start once you can trace an HTTP request, read a small program and use a database; follow each course's prerequisites

Exit evidence: Reproduce failures in independent teaching models, connect them through an order and payment case study, and defend repairs with explicit assumptions and evidence from your own system.

Choose deliberately

Compare the full catalog before choosing your route.

Explore all courses