From data to deployed, secure AI products
The complete path to building AI products: programming and data foundations, machine and deep learning, NLP and large language models, agentic systems, the full web stack around them, and the MLOps and AIOps practices that keep them running.
12 stages · 56 topics · 281 lessons
The tools every later stage is written in and delivered through.
Up next · Stage 02With code, version control and data access in hand, start working with real datasets.
Load, clean, explore and visualise data: the raw material of every model.
Up next · Stage 03Once you can describe data, learn to predict from it.
Classical algorithms and the craft of preparing data, validating and tuning models.
Up next · Stage 04When features are hard to hand-craft (images, text, sequences), let networks learn them.
Neural networks from the perceptron to the transformer.
Up next · Stage 05Apply sequence models and transformers to the richest data of all: language.
Turn text into features and models that understand it.
Up next · Stage 06Large language models collapse most of these tasks into one model. Learn to build on them.
Prompting, chaining, retrieval and tool use with large language models.
Optional · branches from Tools & Agents
Up next · Stage 07When agents need state, control flow and human checkpoints, move from chains to graphs.
Stateful, controllable, multi-agent workflows.
Up next · Stage 08A model is not a product until it has an API, storage and users behind it.
APIs, data storage, identity and reliability around your models.
Optional · branches from Caching, Queues & Testing
Up next · Stage 09Give users an interface to the API you just built.
The web interfaces through which people use AI products.
Up next · Stage 10Now ship it: containerise, automate and deploy repeatably.
Reproducible training, packaged models and automated delivery.
Up next · Stage 11In production, watch the system, and use AI to help operate it.
Observe, evaluate and operate AI systems in production.
Up next · Stage 12Finally, make sure what you ship protects users' data and resists attack.
The security essentials for AI products. The full track is in the Cloud & Cyber Security roadmap.
Real projects built with the skills on this roadmap.