How to use this roadmap
The stages are in order — each one builds on the last. Timings assume around 10–15 hours of focused study a week; go faster or slower as you need. Don't move on until you can complete the stage's checkpoint project on your own.
Stage 1 — Python foundations (weeks 1–4)
- Variables, data types, lists, dictionaries, loops and functions
- Reading files, error handling and writing clean, readable code
- NumPy arrays and the basics of pandas DataFrames
- Checkpoint: load a CSV, clean it and answer five questions about it
Stage 2 — Statistics and probability (weeks 5–7)
- Mean, median, variance, standard deviation and percentiles
- Probability, distributions (normal, binomial) and sampling
- Hypothesis testing, p-values and confidence intervals
- Checkpoint: analyse an A/B test and explain whether the difference is real
Stage 3 — SQL and data wrangling (weeks 8–10)
- SELECT, JOINs, GROUP BY, subqueries, CTEs and window functions
- pandas: merging, grouping, reshaping and handling missing values
- Checkpoint: answer business questions from a multi-table database
Stage 4 — Exploratory analysis and visualisation (weeks 11–12)
- Matplotlib and Seaborn charts; choosing the right chart for the question
- Finding patterns, outliers and relationships
- Communicating findings clearly to non-technical people
- Checkpoint: a short, well-structured EDA report on a public dataset
Stage 5 — Machine learning (weeks 13–18)
- Supervised learning: linear and logistic regression, decision trees, random forests, gradient boosting
- Unsupervised learning: k-means clustering and PCA
- Train/test splits, cross-validation, feature engineering and model evaluation metrics
- Checkpoint: an end-to-end prediction project (for example, customer churn) with a clear write-up
Stage 6 — Deep learning and NLP (weeks 19–22)
- Neural networks, activation functions and training with PyTorch or TensorFlow
- Introduction to computer vision and natural language processing
- Checkpoint: a text classification or image classification project
Stage 7 — Generative AI (weeks 23–25)
- How LLMs work, prompt engineering and calling an LLM API
- Embeddings and retrieval-augmented generation (RAG)
- Checkpoint: a question-answering app over your own documents
Stage 8 — Portfolio and job preparation (weeks 26+)
- Polish 3–4 strong projects on GitHub with clear README files
- Deploy at least one project as a simple web app (Streamlit or FastAPI)
- Practise SQL, statistics, ML and project-walkthrough interview questions
- Tailor your resume and LinkedIn profile to the roles you're applying for
Tips that keep you on track
- Code every day, even for 30 minutes — consistency beats long weekend sessions
- Build projects on data you find interesting; you'll learn more and explain them better
- Learn to read documentation and error messages — it's a core professional skill
- Join a study group or class so you can ask questions and stay accountable