Who's Who in the Data Science Team

The data scientist is often referred to as one profession, although there are other specialties in data science. We will tell you which directions you can choose from and what is required to study them.

When working in the field of data science, you must know the basics and choose one of the narrow specializations. Since data science does not have a universal understanding of them, it will not be easy to do this. Let's try to determine an approximate range of directions that should be guided by. Knowing the opportunities offered by the market will give you a head start in employment, as certain skills are required to get a specific job.

Team roles

The data goes through a series of stages before the decision maker sees a beautiful and understandable presentation. We have arranged positions in the order of their appearance in the process of working with data.

1. Data architect.

Data architecture is the rules, policies, standards and models that define the type of information collected, how it is used and stored. This includes integrating data across an organization and its systems, and addressing security and availability issues. All this is done by the data architect.

Such a specialist is critically needed in big data projects. Usually, in one company, he interacts with several teams at once, sometimes combining his main job with the role of a data engineer.

Typical tasks: along with the development of a detailed plan for the data, the specialist provides the necessary tools and systems for data engineers. In the event of any changes in the company, he understands what will happen to the data and can take measures to minimize the consequences.

What you need to know: SQL, noSQL, XML, Hive, Pig, Hadoop, Spark, machine learning, visualization, data modeling and storage, and ETL (extraction transformation and loan) and a little more.

2. Data engineer.

The knowledge of machine learning and statistics is not required for a data engineer, but it is a very important person on any team. Without them, there will be no data, and therefore no data science.

Typical tasks: These guys are responsible for receiving data, processing it and storing it. They build, test and update the IT infrastructure. Data engineers power everything a data architect has designed. Then the data scientist will have access to information and will be able to run their algorithms.

What you need to know: You will need advanced programming skills to work with large datasets and build the channels through which the data arrives. Technologies: SQL, noSQL, Hive, Pig, Matlab, SAS, Python, Java, Ruby, C ++, Perl, popular APIs and ETL tools.

Perspectives: From Junior to Head of Data Engineering.

3. Data analyst.

This role is less technical than the data scientist, although in many ways they are similar and often confused.

Typical tasks: data analysts answer questions from their colleagues, look for answers among the presented data, perform statistical analysis and translate a bunch of numbers into human language in the form of reports and visualizations. They do not predict or seek new trends on their own. Case Study: Assessing the effectiveness of a marketing campaign and how it affected sales.

What you need to know: Mathematics and statistics, numbers-based decision-making, cleaning (preparation for analysis) and data visualization techniques. You will also need an intermediate level of programming in Python or R, an average level of knowledge of SQL queries, MS Excel, SAS, Tebleau, etc.

Prospects: transition to data scientists if you want to develop in machine learning or data engineers if you are more interested in programming. Data analysts hit the salary ceiling about 10 years after joining the profession.

4. Data scientist.

Due to the lack of clearly delineated roles, some companies are looking for so-called "unicorns": specialists who are well versed in statistics, mathematics, machine learning, programming, business problems and visualization at the same time. There are such people, but there are very few of them.

The average mortal data scientist is simply more immersed in mathematics and programming than the data analyst. He has more freedom to experiment and study trends that management may not be aware of.

The data scientist walks through a sea of ​​unstructured data to identify questions and pull out information that provides the answers. All this needs to be done by understanding the business objectives. And yes, unlike a data analyst, a data scientist is also involved in predictive analysis.

The name of this role includes the word "science" for a reason. There is a process of scientific research - testing hypotheses to gain practical knowledge.

Typical tasks: defining business questions, transforming data, training and tuning machine learning models, evaluating results, predictions, reporting and visualizing by results. Example of work: predicting the likelihood of a customer canceling a subscription, clustering customers by semantic groups.

What you need to know: Scientist data analytics skills + a good understanding of machine learning methods with and without a teacher. It will require a deep understanding of statistics and the ability to evaluate statistical models, as well as more advanced programming skills.

Prospects: Transition as a Lead Data Engineer or Data Architect, Machine Learning Engineer, Lead Data Scientist (Chief Data Officer).

5. Machine Learning Engineer.

In short, a machine learning engineer is (pardon the tautology) a data scientist who specializes in machine learning.

Typical tasks: software solutions for automating ML models; design, development and testing of ML-systems. ML Engineer trains and subsequently maintains machine learning models.

What you need to know: A good understanding of statistics and mathematics is required. Technological stack: Java, Python, JS, as well as ML frameworks TensorFlow or Keras, and, of course, Hadoop or analogs.

6. Business Intelligence (BI) Developer.

The main task of BI developers is to structure and present the obtained data in an understandable form for management. They usually don't do analysis.

It is beneficial to come into this specialty with a non-technical background, because it requires a good understanding of business operations and communication.

Typical tasks: developing strategies for how other employees can effectively use the information received by Scientists and analysts; how to get the necessary information on time to make the appropriate decision. Also, a BI developer designs, creates and maintains repositories, ETL packages, dashboards and analytical reports.

Things to know: SQL, data storage, SSRS / SSAS / SSIS, ETL, Report Builder, Power BI, DAX, Tableau, dashboards, security rules, VB programming languages, C #, JavaScript, etc.

Perspectives: BI developer -> BI analyst -> BI architect -> BI manager.

7. Database administrator (DBA)

When the database is ready, someone needs to look after it. Such a specialist should be able to identify faults, quickly navigate in emergency situations and solve data-related problems.

Typical Tasks: The database administrator is responsible for backing up and restoring information, as well as security and modeling. He makes sure that everyone has access to them, everything works well, and also connects old and new databases.

What you need to know: database languages ​​(most often SQL, NoSQL), as well as the programming language that the company works with. Databases from Oracle and Microsoft, cloud services Microsoft Azure and Amazon Web Services.

Perspectives: manager of computer and information systems.