The AI Platform Engineer Roadmap

A 12-month curriculum for engineers who build production AI systems with coding agents.

Coding agents now write most of the code. This roadmap teaches the rest of the job. You learn to direct agents, check their work, and run AI systems that are reliable, cheap to run, observable, and safe.

• 13 phases
• 12 months
• about 9 hours a week
• 1 real product and 1 capstone

The 13 phases

1. 01Direct coding agents and read their codemonths 0 to 2
2. 02Learn how models behave and what they costmonths 1 to 3
3. 03Build retrieval that finds the right evidencemonths 2 to 4
4. 04Build tool servers with MCPmonths 3 to 5
5. 05Design agent workflows that survive failuremonths 4 to 6
6. 06Run agent workloads on AWSmonths 5 to 7
7. 07Trace every step an agent takesmonths 6 to 8
8. 08Build evals that catch regressionsmonths 7 to 9
9. 09Run agents reliably in productionmonths 8 to 10
10. 10Secure agents and the code they writemonths 9 to 10
11. 11Build an internal AI platformmonths 10 to 12
12. 12Get teams to adopt the platformmonths 10 to 12
13. 13Build Forge, the capstone platformmonths 9 to 12