// the roadmap
The AI Engineer Roadmap
You already know how to code. This is the path from “I can code” to “I can build, ship, and explain real AI systems” — six modules, each ending in a concrete deliverable, free to read.
New here? Start with your role-specific on-ramp.
- ✓Module 01
Concept
The mental model: how LLMs, retrieval, tools, and agents actually fit together.
What you cover
- LLMs, tokens, and context windows
- Embeddings and similarity search
- Structured outputs and function calling
- RAG, tools, and agents — the map
Deliverable
A working script that calls an LLM, generates embeddings, and runs a similarity search.
- ✓Module 02
Build
Turn the concepts into a real retrieval system your users can query.
What you cover
- Chunking and retrieval
- Reranking and grounding
- Citations and function calling
- Adding memory
Deliverable
A production-shaped RAG service and a chat-with-your-docs app.
- ✓Module 03
Harden
Make it trustworthy: measure quality, control cost, and keep it safe.
What you cover
- Building an eval set
- Error analysis and tracing
- Retries, caching, and cost budgets
- Latency and safety
Deliverable
An eval set, tracing, and a cost budget wired into your RAG app.
- ✓Module 04
Deploy
Ship it like software: containers, secrets, serving, and CI.
What you cover
- Containers and secrets
- Model serving
- CI/CD for AI
- LLMOps basics
- ✓Module 05
Explain
Package the work so anyone — including an interviewer — gets it fast.
What you cover
- Architecture diagrams
- Writing a README that sells
- Articulating your trade-offs
- Recording a short demo
Deliverable
A README, an architecture diagram, and a short demo for each project.
- ✓Module 06
Interview
Convert the portfolio into offers: roles, system design, and rehearsed stories.
What you cover
- The role map (AI Engineer, GenAI Developer, Agentic AI Engineer, FDE)
- System design for AI
- Evals and RAG questions
- Project talking points
Deliverable
Rehearsed project talking points and mock system-design answers.
Want the printable version?
Get the roadmap as a PDF plus a production checklist — and Production AI Notes each week.