Roadmap

Three roadmaps, one map each

pick a track

Pick a track and follow its modules in order, or jump to the one you need. 32 modules and 327 topics across AI systems, system design and the maths behind machine learning. Where we have written an explainer the topic links to it, and your reading progress stays in this browser.

hover a module, click to jump

Track 01

AI systems

How modern AI works, from a single neuron to agents in production. One module per chapter of the explainers, in reading order, with a worked architecture where we have one.

Curriculum graph

10 modules in 5 phases

Primary curriculum connection
Parallel or optional connection
Shape, colour and fill distinguish phases

Modules

10 modules, 48 topics

Read in order or jump around. Tick a topic when you have read it. Progress is saved in this browser only, and a tick here is the same tick as on the topic's own page.

01Foundations

2 modules

01

5 topics

Neural network basics

Neurons, layers, loss and how a network learns from its own mistakes.

0 of 5 read0%

02

2 topics

Language as numbers

Text becomes tokens, tokens become vectors, and meaning becomes distance.

0 of 2 read0%

02The model

3 modules

03

8 topics

Inside the transformer

Attention, positions and the residual stream: the architecture behind every modern model.

0 of 8 read0%

04

7 topics

Generating text

From probabilities to words, one token at a time, and the caches that keep it fast.

0 of 7 read0%

05

4 topics

Speed and cost

Experts, quantisation and serving tricks that make large models affordable to run.

0 of 4 read0%

Applied:Model gateway

03Making it good

1 module

06

5 topics

Training and alignment

Pre-training, fine-tuning and teaching a model what people actually want from it.

0 of 5 read0%

Applied:LLM workflow

04Your data and tools

2 modules

07

5 topics

Retrieval and search

Keyword search, vector search, graph indexes and grounding answers in your own data.

0 of 5 read0%

Applied:Hybrid RAG assistant Applied:Agentic graph RAG

08

4 topics

Agents and protocols

Models that plan, call tools, keep memory and talk to your systems through MCP.

0 of 4 read0%

Applied:Single agent with tools Applied:Supervisor multi-agent

05Operate

2 modules

09

5 topics

Observability

How to know a system works, how sure you can be, and how to notice when it stops working.

0 of 5 read0%

Applied:Evaluation loop

10

3 topics

Interpretability and safety

Looking inside the model to see what it is doing, and the failure modes that matter in production.

0 of 3 read0%

Continue learning

Keep going

The explainers are the reading behind the AI track. The topography shows every chapter, topic and term as one connected map. And if you would rather have these ideas working on your data, talk to us.

Explainers Knowledge topography Book a call