Data Scientist Roadmap India: Skills, Stages and Proof of Work
A six-stage data scientist roadmap with the skills to learn and the proof of work to show at each stage, sourced to free learning and documentation pages.
Last updated: 21 September 2026 · By the Asuraa Team
What is the data scientist roadmap?
A data scientist roadmap is an ordered plan for building the skills the role needs and the proof that you have them. The order matters because each stage supports the next: you cannot judge a model without statistics, and you cannot build one without clean data.
If you are new to the role itself, start with our explainer on what a data scientist is. This guide assumes you have decided to try it and want a sequence to follow.
What are the stages of a data scientist roadmap?
There are six stages. The table shows the skills at each stage and the proof of work that shows you have learned them.
| Stage | Skills to build | Proof of work |
|---|---|---|
| 1. Foundations | Spreadsheets, basic maths, thinking with data | A small cleaned dataset with a written summary |
| 2. Statistics | Descriptive statistics, probability, hypothesis testing | A notebook that tests a real question with data |
| 3. Python and SQL | pandas, data cleaning, querying databases | A cleaned dataset and a set of SQL queries |
| 4. Machine learning | Regression, classification, evaluation | A model with a fair evaluation and a plain-language write-up |
| 5. Projects | End-to-end work on real data | Two or three published projects |
| 6. Portfolio and applying | Presenting work, resume, interviews | A portfolio page and a targeted resume |
The stage names and their order are our suggestion. The sources below back the technical content of stages 3 and 4.
What should you learn in the first two stages?
Start with foundations and statistics before code. Comfort with tables, percentages, averages and charts is enough for stage 1, and you can practise it in a spreadsheet.
Stage 2 covers describing data, probability and testing whether a pattern is real. Choose any reputable free statistics course and finish it by analysing a dataset you care about. The proof is a short notebook that asks a question, tests it and states the answer in plain words.
How do you learn Python and SQL for data science?
Python and SQL are the working core of the job. The pandas project's 10 minutes to pandas guide is aimed at new users and covers creating and viewing data, selecting rows and columns, handling missing values, grouping, merging and plotting.
For SQL, Kaggle Learn's Intro to SQL course is described as teaching SQL for working with databases using Google BigQuery.
Practise by combining the two. Pull a table with SQL, clean it in pandas, and write down what you found.
How do you learn machine learning after that?
Move to machine learning once your data skills are steady. Google's Machine Learning Crash Course covers linear and logistic regression, classification, working with numerical and categorical data, neural networks, and real-world topics such as production systems and fairness. It uses animated videos, interactive visualisations and hands-on exercises, and modules are self-contained so you can skip what you already know.
To apply the ideas, use the scikit-learn getting started guide. It covers estimators, pre-processing, pipelines, model evaluation with train-test splits and cross-validation, and parameter searches, and it stresses using pipelines to avoid data leakage.
Evaluation is the habit to build early. A model that scores well on data it has already seen tells you very little.
How do you build projects and a portfolio?
Projects are where the earlier stages come together. Pick a question, find a public dataset, clean it, analyse it, build a simple model if it fits, and explain what a decision-maker should do with the result.
Keep each project small enough to finish. A readme with the question, data source, method, result and limits makes a project easy for a recruiter to read in two minutes.
For the resume, name the tools and results plainly. Our guide to an ATS friendly resume for a data scientist shows how to present this, and the post on data science jobs for freshers helps you decide where to apply.
Where can you get feedback along the way?
Feedback speeds up the roadmap because it catches gaps you cannot see yourself. Ask a practitioner to review one project, and revise it.
You can look for a mentor using our guide on how to find a data science mentor in India. Asuraa also lists mentors, with 1:1 sessions starting from Rs 99, on the mentors page.
What do most data scientist roadmaps get wrong?
Most roadmaps are long lists of courses. These are the gaps that cost learners time.
- They list tools without proof. A certificate shows you finished a course, while a project shows you can use the skill.
- They jump to deep learning too early. The sources above start with data handling, regression and evaluation.
- They ignore SQL. Data sits in databases, and analyst-style roles often ask for SQL first.
- They promise a fixed timeline. Pace depends on your background and practice time, so measure progress by finished projects.
FAQ
How do I start a data scientist roadmap from zero?
Start with basic maths and statistics, then learn Python and SQL. Practise on small real datasets before touching machine learning. The pandas beginner guide and Kaggle's Intro to SQL course are two documented starting points for the Python and SQL stage.
How long does the data scientist roadmap take?
It depends on your starting point, your weekly hours and how much you practise. No reliable source gives one figure for everyone. Set stage-by-stage goals, such as one finished project per stage, and judge progress by proof of work, not by weeks elapsed.
Should I learn SQL or Python first?
Either order works, but do both early. Python handles analysis and modelling, while SQL retrieves data from databases. Kaggle's Intro to SQL course and the pandas beginner guide both give you hands-on practice, so use them side by side in the same weeks.
Do I need machine learning to get my first data job?
Not always. Many first roles, such as data analyst, lean on SQL, Excel or a BI tool, and Python. Machine learning becomes central for data scientist roles, so learn its basics after your data skills are solid and add depth through projects.
How many projects do I need in a data science portfolio?
There is no fixed number. Two or three well-explained projects usually say more than ten shallow ones. Choose projects that show the full path from question to data, method, result and recommendation, and write a short readme for each.
Can I follow this roadmap without a computer science degree?
Yes, the stages are about skills and proof, not a particular degree. Some employers still prefer degrees, so check the postings you want. Build a portfolio, get feedback from practitioners, and apply to analyst-type roles as a way into the field.
Final thoughts
A good roadmap is short on tools and long on evidence. Work through the six stages in order, finish one piece of proof at each stage, and adjust as you learn what the roles you want actually ask for.
To turn this into a personal plan, try the Career Path Planner on asuraa.in.
Related articles
Data Science Internships in India: Skills, Steps and Pay
What data science internships involve in India, the skills employers name, who can apply, seven steps to get one and how to read stipend figures.
Data Scientist Projects: 8 Ideas With Open Data Sources
Eight data scientist projects for freshers, where to find open datasets (including Indian government data), and how to finish and present them.