The skills data scientists use on the job, from programming and statistics through data wrangling, exploratory analysis, inference, machine learning, experimentation, communication, productionising, and LLMs. Start at the top and work down, basics first. Free to read, no signup required.
How to use this: tap a step on the map to open its details, skill pills, and guidance in a side panel. Work down the spine in order; each stage assumes the ones above it. Data science is learned by working with real data, so build projects as you go, because the skills only click once you’ve used them to answer a real question.
Do a real end-to-end data-science project
Show data-science skill with a project taken from messy data all the way to a result you explained. One complete, honest piece of work should anchor your portfolio, and interviewers will dig into it.
Take a real dataset, clean it, explore it, and frame a clear question worth answering.
Build a model with an honest validation setup and a simple baseline, then do proper error analysis.
Write up what you found and what you’d recommend, in plain language for a non-technical reader.
Be explicit about the assumptions and uncertainty, since honesty about limits is a mark of a real data scientist.
Publish it with a clear write-up of your reasoning. The narrative of how you thought about the problem matters as much as the model, because that’s what a hiring manager is really assessing.
Frequently asked questions
Roughly: data analysts focus on answering business questions from existing data with SQL, spreadsheets, and dashboards; data scientists add statistics, experimentation, and machine learning to answer harder questions and build models; ML engineers focus on training models and running them reliably in production. The lines blur and vary by company, and this roadmap covers the science-heavy middle.
More than a data analyst, less than a researcher. You need solid, applied statistics (probability, inference, experiment design) and enough linear algebra and calculus to understand how models work. It’s about sound reasoning under uncertainty far more than advanced proofs.
No, not for most applied roles, though some research-heavy positions favour advanced degrees. A strong grasp of the fundamentals here, plus real projects you can explain, matters more than a specific qualification for the majority of data-science jobs.
Statistics first. It’s tempting to jump to machine learning, but the reasoning that tells you whether a result is real, and whether a model generalises, comes from statistics. Build that foundation and machine learning makes far more sense.
Your starting point matters, and real projects matter more. Someone with a strong quantitative or programming background can move faster. Completing a few end-to-end projects, from messy data to a communicated result, is what speeds it up.
No. Programming, statistics, data wrangling, and machine learning fundamentals are the core. Deep learning, productionising, and LLM work you deepen as your role demands, and data-science jobs vary widely in how much of each they involve.
Get a personalized plan for real interviews
This page covers what to learn. When you’re ready to get hired, Interview Ready turns it into a personalized 30-day plan built around your resume and a specific target role: real practice in the right order (statistics and experimentation, SQL and data wrangling, ML and modelling, product and metrics cases), a guided Build-a-Project track alongside it, and progress tracking the whole way. Start free.
