Most '30-day AI bootcamp' content is designed for people with free time. This roadmap is designed for people without it. The constraint is 5 hours per week. The goal: ship one thing that demonstrates AI engineering skill.
~20 hours
Month 1: Understand + Build Basic
- 2 hours/week reading: what is RAG, what are agents, how LLM inference works (use these lessons).
- 3 hours/week building: get an LLM API key, build a simple chatbot, then add a document (PDF/text) retrieval feature.
- Goal: a working RAG demo.
~20 hours
Month 2: Improve + Evaluate
- 1 hour/week reading: evaluation, guardrails, cost optimization.
- 4 hours/week building: add an eval set (even 20 questions), add basic logging, deploy your project to the web (Vercel, Railway, or similar).
- Goal: a live, evaluatable AI application.
Lessons for this stage
~20 hours
Month 3: Ship + Document
- Polish the project and write a clear README explaining what it does.
- Add it to your GitHub and LinkedIn.
- Write one short post about what you built and what you learned.
- Goal: a public artifact that demonstrates AI engineering competence.
Skill priorities
Learn now
- LLM API + RAG basics
- One working project deployed
- Prompt engineering fundamentals
Learn soon
- Evaluation and basic observability
- Updating your resume/LinkedIn with the project
Learn later
- Agents and complex patterns
- Fine-tuning
- System design at scale
Ignore for now
- Deep theoretical AI
- Building multiple projects before shipping one
- Consuming content without building
- Trying to keep up with every new tool