Zero to AI engineer

Honest timeline

12-18 months of active work is realistic for someone starting from scratch. If someone tells you “become an AI engineer in 3 months” they’re either selling you a bootcamp or selling you an expensive lesson.

Here’s the actual arc for someone who codes 8-10 hours a week and has some programming background but not much:

• Months 1-2: Python fluency + LLM foundations. If Python still feels hard at month 2, spend a third month.
• Months 3-4: Classical ML basics, structured output, tool use, the agent loop. You have your first working AI code.
• Months 5-6: RAG end-to-end. Build one project you’d be willing to demo.
• Months 7-9: Production concerns (evals, cost, gateway), one deep dive into a case file, systems-design fluency.
• Months 10-12: Portfolio projects (2 pieces). Interview prep. Applications.
• Months 12-18: Interview cycle. This takes time. The market is competitive.

Warning signs you’re not ready

• You don’t yet enjoy debugging when things break. AI eng is 80% debugging.
• You want a “recipe.” AI eng has patterns, not recipes.
• You’re studying to escape another career, not to build things. This matters for the interview.

What actually matters at hiring

Not the number of tutorials you’ve done. Not the number of LangChain apps. The three things:

1. One project on GitHub you can demo end-to-end and explain every decision.
2. Ability to reason about failures — evals, cost, latency, security.
3. A minimum-viable engineering practice: reading code, writing tests, using git, deploying something.

If those three feel present, you’re closer than you think.