The Data Scientist path is about extracting insight and drawing sound conclusions from data — framing questions, exploring and understanding data, running experiments, and using statistics and models to answer questions that inform decisions. Where the ML Engineer ships a model as software and the AI Engineer builds on pre-trained models, the Data Scientist’s deliverable is usually an answer, a recommendation, or a communicated finding. Part of the AI/ML course.
What this track covers (and what it borrows)
roadmap.sh’s ai-data-scientist roadmap is deliberately high-level, and much of it — machine learning, deep learning, MLOps, mathematics — is already built in the ML Engineer track and foundations. So this track focuses on the distinctive DS pillars (statistics, EDA, visualization, experimentation, causal inference) and cross-links the shared modeling rather than duplicating it.
Reading order
- The Data Scientist Role — [Beginner] — DS vs ML engineer vs analyst, the analytics spectrum (descriptive→prescriptive), the DS workflow, and why communication is half the job
- Descriptive Statistics — [Beginner → Intermediate] — central tendency, dispersion, distribution shape, and correlation — summarizing data honestly
- Inferential Statistics — [Intermediate] — sampling, the CLT, confidence intervals, hypothesis testing, p-values, and the common tests
- Exploratory Data Analysis — [Intermediate] — the systematic “understand before you model” process
- Data Visualization — [Intermediate] — choosing the right chart, the libraries and BI tools, and communicating without misleading
- Experimentation & A/B Testing — [Intermediate → Advanced] — designing experiments that give trustworthy answers
- Causal Inference & Econometrics — [Advanced] — correlation vs causation, confounders, regression for inference, and time series
What it draws on elsewhere
- Probability & Statistics (foundations) — the probability basis these build on
- Working with Data (ML-engineer) — cleaning/wrangling, shared with this path
- Classical ML (ML-engineer) — the models a data scientist uses, here for insight over deployment
- The ML Toolkit — Python/pandas/scikit-learn (a data scientist also often uses R for statistics)
Related
- AI/ML course map — how the three paths differ
Software engineering
Infrastructure & security
Computer science
Hardware & engineering