AI agent developer

Learn to build, test and run AI agents that use tools, MCP and several models to finish real tasks, working with AI coding tools from week 1, in 20 weeks.

Coming soon · planned, not written yet

20
weeks planned
~40 h
a week
5
milestone projects
1
project, grown every week

These are planned figures. The weeks are written after we check the tools against current job postings and official documentation, so the final route may differ.

Built for the AI era

How AI fits this roadmap

AI coding assistants and agents can now draft much of the glue code in agent projects, so a key skill is directing them, reviewing what they produce and proving it works with tests. This roadmap has you use an AI coding assistant every week and check its output. The agent itself is the subject: tool calling, MCP, orchestration, evaluation and safety limits, because an agent that acts on its own needs much stricter checks than a chatbot.

The main project

TaskPilot

A work assistant agent that reads a request, picks tools through MCP, does the task across several steps, asks a person before risky actions, and is scored by a test set you build.

What you will be able to do

  • 01

    Build a working tool-calling loop in Python against a model API, and handle bad tool arguments, tool errors and loops that never end.

  • 02

    Write an MCP server that exposes your own tools, and connect an agent to it and to an existing MCP server.

  • 03

    Orchestrate a multi-step agent, including one that hands work to a second agent, with saved state and a human approval step before risky actions.

  • 04

    Build an evaluation set with automated scoring, and use it to compare two prompts or two models on the same tasks.

  • 05

    Add tracing, cost and step limits, and defences against prompt injection through tool results, then deploy TaskPilot and write up one failure review.

01The route

Five months, one month at a time.

Each month ends in a milestone project, so you always have something to show for the work.

  1. Month 1 · Weeks 1–4

    Model APIs, tool calling and AI-assisted coding

    Call model APIs from Python, return structured output, and write the tool-calling loop by hand. You set up an AI coding assistant, review its diffs and test what it writes.

    MilestoneA command-line agent with three tools, a step limit and tests, plus a log of AI suggestions you accepted and rejected

  2. Month 2 · Weeks 5–8

    MCP and tool design

    Write MCP servers, design tool descriptions and schemas a model uses correctly, and connect agents to existing servers. You use AI to draft tools, then break them with bad inputs.

    MilestoneAn MCP server with tests, used by TaskPilot over a real connection

  3. Month 3 · Weeks 9–12

    Orchestration, state and several models

    Plan-and-act agents, graph-based orchestration with LangGraph, handoffs between agents with the OpenAI Agents SDK, saved state, approval steps, and routing tasks between models by cost and skill.

    MilestoneA multi-step TaskPilot that pauses for human approval and resumes from saved state

  4. Month 4 · Weeks 13–16

    Evaluation, tracing and safety

    Build test sets, score runs automatically, trace every step, and cap cost and steps. You test for prompt injection through tool results and limit what each tool is allowed to do.

    MilestoneAn evaluation report comparing two setups of TaskPilot, with traces and a list of attacks it now blocks

  5. Month 5 · Weeks 17–20

    Deployment, capstone and job search

    Run TaskPilot as a service with logs, budgets and alerts, review one real failure, write a portfolio case study, prepare for interviews and apply.

    MilestoneA deployed agent, one failure review, a portfolio write-up and a working job-search pipeline

03What is inside

Everything a week comes with.

Every week is built the same way, so once you have done one you know how all of them work.

  • 01

    One project, TaskPilot

    You grow one agent from a loop on your laptop to a deployed, tested service, so every tool lands in something real.

  • 02

    A day-by-day plan

    Each week has a Learn part and a Do part, with a checklist so you know when you are done. Most weeks include a task to do with an AI assistant and a step to check its work.

  • 03

    A project in every milestone

    Five milestones, each ending in something you can show and link to.

  • 04

    Reference topics

    Short topic pages with check questions you can come back to.

  • 05

    A job track

    Interview practice, portfolio and resume help, and an application plan.

  • 06

    Progress in your Study plan

    Add weeks to your Study plan, give them dates, and finished days are counted for you.

04Is it for you

Who this is, and is not, for.

Better to know before you start than six weeks in.

It is for you if

  • You can already write working Python or TypeScript and want to build agents, not just call a chat API.
  • You are a backend, full-stack or data developer who wants to be strong with AI tools and with building on them.
  • You can give about 40 hours a week for five months.

It is not for you if

  • You have never written code. Do a beginner programming course first.
  • You want a job guarantee. Nobody can promise that, and this roadmap does not.
  • You want to train models from scratch. This roadmap builds on existing models.
  • You want to paste AI output into production without reading it. This roadmap trains you to check it.

?Questions

Before you start

Is this roadmap ready?
Not yet. This is the plan. The weeks are being written and checked against current job postings and official documentation. Leave your email and we will write to you when it opens.
What will it cost?
The price is not set. Roadmaps are bought once, with no subscription. We will announce it when this one opens.
Will the plan change?
It may. Before the weeks are written we check the tools against current job postings, so the lengths and months here are planned, not final. Agent tools change fast, so expect the framework names to be reviewed again.
Do I need a paid model account?
Probably a small one. Calling model APIs costs money per use. The roadmap will show how to cap spend and use cheaper models for most practice, and will say which parts can run on free tiers once the weeks are written.
Will it get me a job?
It will give you skills and proof of work. It cannot promise a job on a fixed date, because hiring depends on the market, your location and your past work.

Not open yet

Leave your email and we will write once, when it opens.

Notify meAll roadmaps