All learning paths

Course outline

Intermediate

Explore transformers and build applications with retrieval, model APIs, tools, and agents.

44 lessons9 modulesAbout 22 hours with practice

01

Transformer basics

Follow attention, encoders, decoders, and position information through a transformer.

1A short history of NLP25 min2How transformers work29 min3How does self-attention work?33 min4Encoders and decoders25 min5How transformers represent position27 min6T5 and text-to-text learning27 min
02

Language model architectures

Compare GPT, BERT, reasoning models, and the tasks different model designs support.

1From GPT to chat assistants27 min2BERT and masked language modeling27 min3How do reasoning models work?27 min4How decoder-only models generate text28 min5Using encoder-only models27 min
03

Applications with LangChain

Make model calls, connect tools, manage conversation memory, and build a small agent.

1Getting started with LangChain27 min2Chains, agents, and tools28 min3How conversation memory works30 min4Build a question-answering app with LangChain35 min5Build a multi-step research agent40 min
04

Retrieval-augmented generation

Find relevant documents and use them to answer questions with sources.

1What is RAG?31 min2What is a vector database?28 min3How to measure embedding similarity28 min4Build your first RAG pipeline35 min5Build a document question-answering app40 min
05

Embeddings and search

Choose embeddings, split documents, compare vector stores, and build search.

1Choosing embeddings for search28 min2How to compare vector databases30 min3How to split documents into chunks30 min4How hybrid search works30 min5Build a semantic search engine40 min
06

Agent frameworks

Build small agents with Google ADK, the OpenAI Agents SDK, and CrewAI.

1Inside an AI agent's control loop25 min2Getting started with Google ADK30 min3Getting started with the OpenAI Agents SDK30 min4Organizing agent tasks with CrewAI30 min5Compare agent frameworks on one task40 min
07

MCP tools and workflows

Connect tools with MCP and build workflows that track their state and recover from interruptions.

1How MCP connects tools to an assistant24 min2Build a read-only MCP server30 min3Designing AI workflows25 min4Building workflows with LangGraph30 min5Build a workflow that can resume40 min
08

Tools for AI application development

Use MCP, LlamaIndex, and the Vercel AI SDK in applications with user accounts.

1The MCP request lifecycle25 min2Getting started with LlamaIndex28 min3Streaming responses with the Vercel AI SDK30 min4Reviewing code from AI tools25 min5Build an AI app with user accounts45 min
09

Testing AI applications

Validate structured output, select useful context, and evaluate results.

1Structured outputs and JSON Schema31 min2What is context engineering?28 min3How to evaluate a language model30 min
Take what you learned into a challenge