Who is a Data Scientist, what do they do, and how much do they earn?

Data scientists are needed in virtually every industry, from manufacturing to streaming services. For example, in retail, a data scientist analyzes data on customer behavior in-store → creates a model for optimal pricing → increases the average order value → generates profit, and the company makes a profit.

The essence of a data scientist’s work is to use algorithms that have already been developed and to understand which ones to use and when. For example, Netflix’s AI, which recommends TV shows and movies based on what people like and watch, is also the result of a data scientist’s work.

1. Clarify the requirements for a business problem and translate it into mathematical terms.
2. Prepare the data to solve the problem: figure out where to get it and how to process it so that it becomes available for work.
3. Analyze and structure data.
4. Build a machine learning model that will solve the problem.
5. Check that the model works correctly: implement it on a set of users or conduct A/B testing.

This list can be looped and returned to the data collection or model training point if the current one fails.

For example, a client wants to increase revenue from marketing emails. To solve this problem, a data scientist must first understand which metrics impact revenue.

To do this, he will ask marketers for newsletter data stored in a database or Excel spreadsheet. A data science specialist will compile this data and segment the newsletter recipients into those who accepted the offer and those who did not.

Next, the Data Scientist evaluates whether there is enough data to build models, and if so, writes an algorithm that will send each subscriber a tailored email.

After that, all that’s left to do is test the mailing on a small number of users and measure its effectiveness. If it’s higher, you can celebrate your success. If not, you’ll have to return to the data collection stage and repeat the entire process.

Data scientists are often confused with data analysts because their tasks seem similar at first glance. Both work with large data sets and have excellent knowledge of their domains, such as markets and industries, but there are subtleties.

An analyst’s job is to conduct statistical analysis to answer questions or solve problems. To do this, they collect data, identify patterns, and generate reports that help project or business managers make strategic decisions.

A data scientist can not only analyze and visualize data but also build models based on it. This requires knowledge of machine learning and deep learning, which analysts lack.

An ML engineer (Machine Learning Engineer) continues the work of a Data Science specialist if she has shown good results.

A data scientist analyzes data, builds models, and tests them. An ML engineer’s responsibilities include automating models, ensuring they perform well, and troubleshooting errors. If a model’s accuracy drops, the engineer will investigate the cause and retrain the algorithm.

We’ve compiled a list of the key tools a data scientist uses.

Programming languages

The primary language for data scientists is Python: it’s convenient for data analysis, sample preparation, and model building. SQL and ClickHouse are used for database queries, and R can be used for research tasks in certain fields.

Libraries and frameworks

• Pandas and NumPy are basic data processing tools.
• Matplotlib, Seaborn, Plotly – visualization.
• scikit-learn — classic machine learning algorithms.
• PyTorch is the primary framework for deep learning and neural network development in Russian companies.
• TensorFlow is used less frequently, but is found in projects with an existing codebase or industrial ML.
• spaCy, NLTK, and models from Hugging Face for working with text.

Development environments and version control

For experiments and quick hypothesis testing, data scientists typically use Jupyter Notebook. For larger projects, they use PyCharm or VS Code.

Teamwork is most often carried out through Git and repositories in GitLab or GitHub.

Databases and storage

In addition to PostgreSQL and ClickHouse, MySQL and MongoDB are also used. Working with large volumes of data requires distributed processing tools, such as Apache Spark. Large corporations and government agencies can host data and computations in Yandex Cloud or VK Cloud.

Implementation of models

To make the model work in the product, it is packaged using Docker and connected via FastAPI or Flask. MLflow, Airflow, and DVC are used for quality monitoring and scheduled model training.

At the start of your career, mastering the basics is sufficient: Python, Pandas, scikit-learn, SQL, and Jupyter Notebook. Additional tools are added as tasks become more complex, and the data scientist advances in their profession.

To work as a data scientist, you need two types of skills: technical and cross-functional. The former are related to specialized disciplines, while the latter relate to psychological qualities and management skills and are essential for any specialist, regardless of their profession.

The skill distribution of a Data Scientist is heavily skewed towards technical skills because most of their work involves dealing with data rather than people.

• Programming in Python and SQL.
• Mathematics, statistics, and machine learning.
• Working with databases.
• Proficiency in big data processing tools: Apache Spark and Hadoop MapReduce.
• Productivization of models.
• Advanced Proficiency English for reading technical literature.
• Understanding the specifics of the business and the domain area.

• Communication with colleagues.
• Presentation of the results of your work.

1. Basic knowledge of machine learning and statistics. Understanding of key algorithms and their applications.
2. Experience: not required, maximum – a training project.
3. Programming: Confident in Python and basic knowledge of SQL.

1. Deep knowledge of mathematics.
2. Experience: 2-3 completed projects.
3. Programming: Confident Python skills and knowledge of its features in terms of model productivity and work optimization.
4. Strong knowledge of experimentation culture and working with tools for implementing and supporting machine learning models: GitLFS, MLFlow, DVC. Knowledge of A/B testing.
5. Ability to solve a problem from start to finish with minimal intervention from a senior specialist or team leader.

1. Deep, confident knowledge of mathematics and statistics.
2. Experience: from 5 completed projects.
3. Programming: confident Python, SQL.
4. Expert knowledge in your field.
5. Complete independence from setting the task to putting it into production.
6. Ability to train and mentor junior and advanced professionals.

In recent years, the demand for data scientists has only grown. All major companies are opening data science departments. Startups and small development teams also need specialists.

New problems that can be solved with data science are constantly emerging. Modern machine learning models help us solve even year-old problems in new ways—and earn more.

The path of a data scientist is one of continuous professional development. The tasks for data scientists are becoming more complex and interesting. For example, creating support chatbots and voice assistants using NLP (Natural Language Processing) or machine learning based on text data.

Graduated from a relevant university and the machine learning department at HSE, MIPT, or Moscow State University. Studying as an applied mathematician at a non-research university is also an option.

1. A data scientist is a specialist who works with data to solve business problems. They work at the intersection of programming, machine learning, and mathematics.
2. A data scientist’s primary responsibilities include collecting and analyzing data, building models, training them, and testing them. They must understand how the company operates and the specific industry they work in.
3. The data scientist profession is constantly evolving and highly paid. New and exciting challenges are constantly emerging. The demand for data scientists at large companies will only increase, as will their salaries.
4. To become a data scientist, you don’t necessarily need to graduate from a specialized mathematics university. You can get additional education, take an internship, or get a job as a junior specialist.

• A Data Analyst works with existing data: creates reports, dashboards, prepares visualizations, and performs basic analytics.
• A Data Engineer creates data infrastructure (databases, pipelines, processing large volumes of information).
• A data scientist combines analysis, programming, and statistics skills to build models, forecasts, and algorithms based on data.

Data Scientist remains one of the most in-demand professions in 2025: over the past four years, the number of vacancies in this field has grown 2.5 times. At the same time, the role is expanding: new areas of expertise are emerging in MLOps, generative AI, and model implementation, with data scientists participating in business decision-making, not just analyzing data. Furthermore, the profession is expanding into new industries, including healthcare, manufacturing, energy, and government.