Skip to content
NextiaNextia Learning

Learning path · Build AI

Agent and MCP application engineer

Build AI agents that use tools and data safely, with MCP.

This is you

You want to build AI that does work, not only AI that talks: agents that call tools, read data and finish tasks. You can write some Python, or you will learn it here.

Where you are going

An agent and MCP application engineer

  • Choose between a fixed workflow and an agent, for the right reasons
  • Give an agent tools and data through MCP, with limits
  • Add human approval where a mistake would cost a lot
  • Secure, test and deploy an agent for real users
  • 3 stages
  • 14 courses, 14 available now
  • About 202–274 hours of study
  • 14 certificates of completion

How much code you need

  1. No code
  2. Read code
  3. Some code
  4. Code is central

Code is central: you build agents, servers and their tests. The lessons go one small step at a time.

Your route, stage by stage

Each stop is a course. Its shape tells you how much of it to take.

  • Full course. The whole course, with its practice. It leads to a certificate of completion.
  • Concepts. Only the concept modules, read without running code. They lead to a concepts certificate.
  • Optional. Useful for some people. You do not need it, and it is not counted in the time.
  1. You, today

  2. Stage 1 Foundationsabout 46–62 h

    1. AI Career Foundations: Roles, Skills and Roadmaps

      Full course4–6 h

      See where agent work fits in AI teams.

    2. Python for Practical AI Engineering

      Full course18–24 h

      Python for every agent that you build.

    3. Essential Mathematics and Statistics for AI

      Full course12–16 h

      The numbers behind agent results.

      Focus: If you already studied this, do its final assignment instead.

    4. SQL and Data Preparation for AI

      Full course12–16 h

      The data that agents read and change.

  3. Stage 2 LLMs and qualityabout 50–68 h

    1. LLMs: How They Work and How to Choose One

      Full course8–12 h

      How LLMs work, and how to choose one.

    2. Building Reliable Applications with LLM APIs

      Full course22–30 h

      Tool calling starts here.

    3. Evaluating and Testing AI Systems

      Full course20–26 h

      Test agents before you trust them.

  4. Stage 3 Agents in productionabout 106–144 h

    1. AI Agents and Workflow Orchestration

      Full course22–30 h

      The core of this path: agents and workflows.

    2. Build and Deploy Your First API

      Full course11–15 h

      Serve your agent behind an API.

    3. Docker and Cloud Essentials for AI Developers

      Full course7–10 h

      Run it in a container, in the cloud.

    4. Databases and Storage for AI Applications

      Full course16–22 h

      Store state, history and documents.

    5. Authentication, Permissions and Multi-Tenant Applications

      Full course16–22 h

      Agents act for a user: check who, and what they may do.

    6. AI Security, Privacy and Responsible Design

      Full course18–24 h

      Stop prompt injection and unsafe tool use.

    7. MCP: Connect AI Applications to Tools and Data

      Full course16–21 h

      Connect agents to tools and data with MCP.

  5. You arriveAn agent and MCP application engineer

Earn certificates as you go

You do not need to finish the whole path to show what you learned. Each full course gives a certificate of completion, and each concepts course gives a concepts certificate. Share each one when you get it.

When you are ready, prepare for the interview: Interview preparation: AI Engineer

Make it my path All learning paths

Choose this goal in your profile. My learning then shows your progress on this route, and the weeks it takes at your pace.