A data scientist turns messy data into decisions. You find a business problem (why are customers leaving? which leads will buy?), pull and clean the data, build and test a model or analysis, then explain the result so someone can act on it. Most of the job is not fancy algorithms: it is SQL, cleaning data, careful statistics, and clear communication. This roadmap builds in order: Python and SQL, then statistics, then machine learning, then deployment and leadership. Every resource is free. Work at about 8-10 hours a week, build a project at each stage, and publish it on GitHub. Hiring managers
See the full roadmap free
Every stage from beginner to expert, with lessons, practice drills, projects, interview questions and free resources. Free with a Rolearch account.
Stage 1 of 4 · about 14 weeks
Learn Python (variables, functions, pandas, NumPy), basic SQL (SELECT, WHERE, GROUP BY, JOIN), Git, and core statistics (mean, variance, distributions, sampling, confidence intervals, hypothesis tests). Finish by cleaning a public dataset and writing up 3 findings with charts.
Python for data work
Python is the main language data scientists use to clean, analyze and model data. Every later stage, from machine learning to pipelines, is built on being comfortable writing small, clear programs.
Data wrangling with NumPy and pandas
Most of a data scientist's time goes into getting messy raw data into a clean, trustworthy table. The core task list says it plainly: clean and manipulate raw data, and analyze large sets with statistical software. Models are only as good as this step.
Basic SQL
Company data lives in databases, not CSV files. Nearly every data scientist job posting lists SQL, and interviews test it. Even the software list for this role includes Redshift and Hive, which are queried with SQL-style language.
Git and GitHub for reproducible work
Your portfolio and your teamwork both live in version control. Employers read your GitHub, and in a team you will share code, review changes, and need to recover from mistakes.
Core statistics for data analysis
Statistics tells you whether a pattern is real or just noise. Without it you will report differences that vanish next month. The role depends on mathematics, and later modeling stages assume you understand these ideas.
Then 3 more stages (about 102 weeks in all) with lessons, drills, projects, 10 interview questions with strong answers and free resources.