About 100 hours

Agentic AI engineering roadmap.

A focused course of about 100 hours for software engineers preparing to build AI applications. Start with LLM fundamentals, then move through retrieval, tools, agentic workflows, and evaluation before completing the Customer Service Agent capstone.

7 phasesAbout 100 hours1 capstone project
02

Phase 02

Prompt Engineering, Context Engineering, and Structured Outputs

12 hours

Design model requests, assemble trusted context, and return validated data for application code.

03

Phase 03

Retrieval-Augmented Generation (RAG)

18 hours

Retrieve relevant, current information before an LLM writes an answer.

04

Phase 04

Tool Calling and Model Context Protocol (MCP)

15 hours

Connect an LLM to tools while application code keeps control of permissions and execution.

05

Phase 05

Agentic Workflows

17 hours

Design agent workflows with state, controlled decisions, and human oversight where needed.

06

Phase 06

Evaluation, Observability, and Safety

18 hours

Evaluate AI behavior, trace requests, protect users and data, and manage cost and releases.

07

Phase 07

Customer Service Agent Capstone

25 hours

Bring the roadmap together in a small public application for customer refund requests.

Plan for around 12 to 13 hours each week to complete the course in two months. The estimate includes reading, implementation, debugging, evaluation, and capstone work.