AI Engineering Roadmap
A learning route through the AISL material: what to learn first, which course or workshop covers it, and what to build along the way.
The material does not have to be consumed in publication order. This page stitches it into a route: foundations, then retrieval, then agents, then production concerns, with a portfolio project as the constant companion.
Stage 0: Python
Everything else assumes you can write Python comfortably. Python for AI Engineering builds a podcast aggregator from a notebook into a structured command-line application: fundamentals, functions and code organization, external data, pandas, errors and testing, and object-oriented Python.
Stage 1: The full arc, once
AI Hero: 7-Day AI Agents Crash-Course is the fastest complete pass: ingest a real GitHub repository, chunk and index it, build a RAG assistant with tools and citations, evaluate the answers, and ship it behind a Streamlit interface. One week, end to end, portfolio-ready.
Stage 2: Retrieval properly
After AI Hero, go deeper on the part most demos get wrong:
- RAG — the pipeline, and why fixed retrieval breaks
- Vector Search — TF-IDF to embeddings, built from scratch in Build Your Own Search Engine
Stage 3: Agents
- Start with the mechanism: The Agentic Loop from Scratch
- Then frameworks and protocols: Building AI Agents with MCP, PydanticAI, and OpenAI
- The AI Engineering Buildcamp covers this stage — and retrieval and production — as a redesigned curriculum: nine teaching weeks plus optional hackathon weeks, with homework, capstones, and two peer-reviewed project attempts (see the course record for the week-by-week outcomes).
Stage 4: Production
Deployment, guardrails, testing, monitoring, and evaluation have their own pages: Agent Deployment, Guardrails, Evaluation and Monitoring.
Stage 5: Ship a portfolio
The project is the point. Selecting a Portfolio Project gives a framework for aiming projects at the job you want — the Portfolio Projects page reads the site's ten showcase projects against it — and the sprints described in AI Shipping Labs Sprint 2 Starts Today give you six weeks of structure and company to finish one.
Where to go next
- Check the role itself: AI Engineering
- Career mechanics — CV, brand, interviews: AI Engineering Career, and the interview itself in Interview Prep
- The same route recut for your starting point — backend, data engineer, data scientist, frontend, or ML engineer — with the market data behind the ordering: Field Guide: Learning Paths
Keep building with a membership
Every topic guide is free to read. Members get structured courses, live workshops, and feedback on real projects.