Career Roadmap

AI and Data Scientist

Data Science turns raw numbers into decisions — and increasingly, into the models and AI systems that make decisions on their own. This roadmap walks you from the math and statistics foundation through coding, exploratory analysis, classic and deep machine learning, MLOps, and the AI engineering and vibe coding skills reshaping how data scientists work today.

Why AI and Data Science?

The discipline behind every "the data shows..." claim a company makes

Every company sitting on data wants the same thing: to turn it into decisions, predictions, or products. That's the job of a data scientist — running the experiment that tells you if a feature actually helped, building the model that predicts churn before it happens, or shipping the LLM-powered feature that used to sound like science fiction. It's a field that rewards rigor (a wrong A/B test conclusion is worse than no test at all) as much as it rewards curiosity, and it now stretches from classical statistics all the way to prompting and fine-tuning large language models.

You don't need a specific degree to get in — Computer Science, Math, Statistics, Physics, and Economics all provide a strong base, and plenty of practitioners have transitioned in through bootcamps and self-study, since practical skill matters more than the credential. The field also branches into a few closely related roles. A Data Scientist focuses on statistical rigor, experimentation, and classical/deep ML. An AI Engineer leans further into building and deploying LLM-powered applications. A Data Analyst focuses more on reporting, dashboards, and business-facing insight. Whichever direction you lean toward, the roadmap below covers the shared foundation all three draw on.

Quick intro — what does a data scientist do?

A quick primer before you start the roadmap. Opens in a small player, no need to leave the page.

The AI and Data Scientist Roadmap

Work through these in order. Each step has a short lesson, reference reading, and the tools data scientists rely on daily.

STEP 1

Mathematics & Statistics

Linear algebra, calculus, and mathematical analysis, followed by the statistics that everything downstream depends on: distributions, CLT, and hypothesis testing.

Documentation
STEP 2

Experimentation & A/B Testing

Experiment design, minimum detectable effect, and variance-reduction techniques like CUPED that make your tests sensitive enough to trust.

Documentation
STEP 3

Econometrics & Time Series

Regression fundamentals, fitting distributions, and time series modeling — ARIMA and forecasting techniques for data that moves through time.

Documentation
STEP 4
Python logo

Coding: Python & SQL

Python programming, data structures and algorithms, and SQL — the query language every dataset eventually passes through.

Documentation
STEP 5
Pandas logo

Exploratory Data Analysis

Understand a dataset before modeling it: cleaning, visualization, and exploration with Pandas and Seaborn.

Documentation
STEP 6
scikit-learn logo

Machine Learning

Supervised and unsupervised learning, advanced techniques like ensembles, and enough theory to know why a model works, not just that it does.

Documentation
STEP 7
PyTorch logo TensorFlow logo

Deep Learning

Fully connected networks, CNNs, RNNs/LSTMs, and Transformers — plus transfer learning to build on models others already trained.

Documentation
STEP 8
MLflow logo Docker logo

MLOps

Get models out of the notebook and into production: deployment patterns, model monitoring, and CI/CD for machine learning.

Documentation
STEP 9
OpenAI logo Hugging Face logo

AI Engineering

Prompt engineering, working with LLMs, retrieval-augmented generation (RAG), agents, and fine-tuning to adapt models to your data.

Documentation
STEP 10

Vibe Coding

AI coding assistants and AI app builders that let you prototype data tools and pipelines faster than writing every line by hand.

Documentation

Practice Projects

Real, buildable projects to put on your own GitHub or Kaggle profile

Track complete

Ten steps, from linear algebra to LLM agents. Push your notebooks and models to GitHub or Kaggle so they're visible to employers, then keep going — data science is learned by running real experiments, not just reading about them.

Where next?

Keep exploring by domain or drill into a single skill

Video