{}Route To AI
Roadmaps
← All roadmaps
Intermediate

Become an AI agent engineer

From how models work to agents you can ship and trust.

For: Developers who can already code and want to build with LLMs and agents.

5 steps
stops on the route
13 lessons
you finish in total
2 courses
the lessons come from
~4.5h reading
all lessons together
Adds it to My learning so you can track it.
AI careerThis is the roadmap for the Agentic engineer role.
Read about the role →

Finish every lesson to earn a certificate of completion that anyone can verify. Jump to the certificate ↓

Other roles:Agentic orchestratorForward deployed engineerAI transformation strategistAI evaluation engineerAI security engineerAI governance and risk leadAI product manager
Your progress on this roadmap0 / 13 lessons completed0%

How to read this page. A roadmap is a route made of 5 steps. Each step has one or more lessons. There are 13 lessons in total, taken from 2 courses. Extra reading is listed at the end of the page. It does not count towards your progress.

Certificate of completion

Earn your Agentic engineer certificate

Verified edition

Mark every lesson in this roadmap complete. You have finished 0 of 13. 13 to go.

See a sample certificate. Anyone can check a certificate with its code at /verify. It records course completion. It is not an exam or an accredited qualification.

  1. 1

    Step 1 of 5

    Understand the model

    Explain how an LLM works, and what it can and cannot be trusted with.

    Lessons0/2 done
    Tool: Hugging FaceTool: Ollama

    Before you move on: Get structured JSON back from a model and handle it when the format is wrong.

  2. 2

    Step 2 of 5

    Give it tools and knowledge

    Let a model call functions and answer from your own data.

    Lessons0/2 done
    Tool: ChromaTool: pgvectorTool: Pinecone

    Before you move on: Build a small question-answering app over ten of your own documents.

  3. 3

    Step 3 of 5

    Build the agent

    Choose a framework (or none) and build a multi-step agent with memory.

    Lessons0/3 done
    Prompts: AI agent promptsTool: Model Context Protocol (MCP)Tool: LangChain & LangGraphTool: OpenAI Agents SDK

    Before you move on: Write the same simple agent twice: once by hand, once with a framework. Compare them.

  4. 4

    Step 4 of 5

    Test it and ship it

    Measure quality, find failures, and deploy with monitoring.

    Lessons0/2 done
    Tool: LangfuseTool: Braintrust

    Before you move on: Make a test set of 20 real examples and run it before every change.

  5. 5

    Step 5 of 5

    Go deep as an architect

    Design whole systems: security, cost, cloud and trade-offs.

    Lessons0/4 done
    Tool: Amazon BedrockTool: Cloudflare Workers AI

    Before you move on: Draw the design for one system and explain each trade-off out loud.

Hands-on lab · build confidence

Build a support agent with two tools and a safety limit

A small agent that looks up orders and opens tickets, with checks, limits and a test set.

Some code · 4 to 6 hours

Start the lab →Practice interview questions

Other roadmaps

Use AI well at work

Get reliable results from AI in your daily job, with no coding.

Design and run multi-agent systems

Know when one agent is enough, and how to run many safely.

Design and ship websites with AI

Go from a brief to a live, good-looking site, and work with clients.

Create images, 3D and video with AI

Make a consistent, usable visual style, and know the licensing rules.

0 / 13 finished0%