Most people learn like this:
Course → Course → Course → “Now I’m ready to build something”
We’re seeing the fastest learners flip it:
Project → Learn what’s needed → Project → Learn what’s needed
They learn through building, not before building
Sounds obvious. But watch what happens in practice:
Someone wants to learn AI. They take a course on Python. Then a course on LLMs. Then a course on Agentic AI. Then a course on AWS Deployment.
Six months later, they’ve “learned” a lot. Built nothing. Can’t prove anything.
Meanwhile, someone else picks a project on day one. A simple one: predict something, classify something, automate something.
They get stuck immediately. They don’t know how to clean the data. So they learn that. They don’t know which model to use. So they learn that.
Every gap in knowledge becomes obvious… and gets filled.
Six months later, they have a portfolio. They have stories to tell. They have proof.
Same amount of time. Completely different outcome.
The shift:
Stop preparing to start.
Start to prepare.
Pick a project that’s just beyond your current skill. Figure it out as you go. That’s how the fastest learners actually learn.
Not sure where to start?
- Build a chatbot that answers questions about a PDF (LLMs)
- Automate a weekly report you do manually at work (automation)
- Scrape job postings and analyze what skills are most in demand (scraping + analysis)
- Take a pre-trained ML model and wrap it into a production-ready API (MLOps)
- Build a dashboard of your 2025 Spotify listening history (data viz)
Want something more advanced? Try an AI-powered Travel Planner challenge → GitHub repo.
Pick one. Start today. Learn what you need as you go.
Switch to builder-first mode,
Kirill
P.S. Courses aren’t the enemy. Using them to avoid building is.