The single biggest reason learners stall in AI isn’t talent — it’s the lack of a real roadmap. Tutorials are infinite; a roadmap is finite. This is the six-month plan we’ve seen work for people coming from adjacent fields.
Months 1–2: Foundations
Python, math refresher, statistics, and SQL. Pair every week with one tiny exercise on Kaggle so the abstract math meets real data.
Months 3–4: First project
Pick a problem you care about. Use classical ML before you reach for deep learning. Ship the project with a README, a clean repo, and a one-paragraph write-up.
Months 5–6: Specialize and apply
Pick a niche — LLMs, computer vision, ML systems — and double down. Add one more portfolio project in your niche and start applying.
The bottom line
Roadmaps fail when they’re too detailed and when they’re too vague. The version above is small enough to fit on one page and concrete enough to start tomorrow.
