Why most "learn AI" plans fail
Beginners usually drown in tutorials — three courses on Python, two on math, a YouTube series on neural networks — and never build anything that works end-to-end. The fix is the opposite: learn the smallest amount of theory needed to ship the next project, then repeat.
This roadmap takes about 6 months at 1–2 focused hours per day. Each month ends with a project you can put on GitHub.
Month 1 — Python Foundations
- Variables, data types, control flow, functions
- Lists, dicts, comprehensions, file I/O
- NumPy and Pandas basics for data handling
- Build: a CSV analyzer or simple scraper
Month 2 — Math You Actually Need
- Linear algebra: vectors, matrices, dot products
- Calculus intuition: gradients and derivatives
- Probability & statistics: distributions, Bayes
- Build: implement linear regression from scratch
Month 3 — Classical Machine Learning
- Supervised vs unsupervised learning
- Regression, classification, clustering
- scikit-learn workflow: train, validate, evaluate
- Build: end-to-end Kaggle starter project
Month 4 — Deep Learning
- Neural networks, activation, backpropagation
- PyTorch (or TensorFlow) fundamentals
- CNNs for vision, RNNs/Transformers for sequences
- Build: image classifier on a real dataset
Month 5 — Modern AI & LLMs
- How Transformers and LLMs actually work
- Prompting, embeddings, fine-tuning basics
- RAG: retrieval-augmented generation
- Build: a chatbot over your own documents
Month 6 — Portfolio & Career
- Ship 2–3 polished projects to GitHub
- Write up each project with results & tradeoffs
- Contribute to open source, apply to internships
- Build: a portfolio site linking everything
Frequently asked questions
Do I need a math degree to learn AI?
No. You need working intuition for linear algebra, calculus, and probability — not graduate-level theory. Learn the math alongside the code so it stays concrete.
Should I learn machine learning or jump straight to LLMs?
Spend at least a month on classical ML first. Concepts like overfitting, train/test splits, and evaluation metrics carry over to every modern AI system, including LLMs.
How long does it take to learn AI from scratch?
With consistent daily practice, most beginners reach a hireable junior level in 6–9 months. Pathlix generates a personalized week-by-week plan so you always know what to do next.
Is Python the only language I need?
Python is enough for 95% of AI work. Add SQL for data and a little JavaScript if you want to ship AI features into web apps.
Get this roadmap personalized for you
Pathlix turns this 6-month guide into a week-by-week plan tailored to your background, target role, and available study time — with daily goals and weekly checkpoints.