Building Data Science Team: How to Handpick an Effective Crew
If you are planning to build a big data project team this article will definitely come in handy because today we are talking about the most effective tips & tricks for hiring professional specialists that will boost your business success.
The ironic thing about recruitment is that it is similar to data analysis process: you have to “filter” all candidates and find those, who have competences and skills that perfectly fit your corporate strategy and can fully attain the objectives set. According to Statista, 42% of big data companies state that their biggest challenge is maintaining data quality. That is why it is so important to hire experienced people, who will ensure the highest work quality.
Types of Data Science Projects You Can Encounter at Your Work
For building a data science team, in which each member has a certain set of skills, you need to take into account details of the project they will work on. Two different specialists can be professionals in different domains but it does not mean they are both equally good. Therefore, the main task is to choose the one, who has the skills applicable for your particular project.
Moreover, you have to foresee the possibility of emergence of new projects you may launch in the nearest future because replacing employees is not like replacing office equipment. Human resources are much harder to find and allocate. That is why your employees should have strong knowledge in the domains you consider promising.
General types of Data Science (DS) projects are:
- Data visualisation;
- Data processing;
- Machine learning;
- Data cleaning;
- Streaming data.
Taking into account the types of project you are going to launch or consider for the future, you can define which skills are a must for your team.
What Data Science Skills a Perfect Candidate Has to Possess
The first questions every CEO, CTO, or HR asks oneself is “How many members do I need to have in my DS team? How many DS engineers to hire? How many data scientists do I need?”. To be completely honest, it’s not about the quantity, it’s about building a big data team with the right structure. But we are ready to provide you with answers to those questions too.
For a company consisting of 300 employees, 5-8 persons will be enough. Big data team structure can differ depending on the project complexity and deadlines set. Though, for typical tasks, a forementioned number is the optimal one.
Data science team structure:
- 1-2 data scientists;
- 3-5 data engineers;
- 1 data science team lead.
Data Scientist Skills
DS is a candidate with a deep and solid quantitative background. As for education, this person should have a Ph.D. in either computer science/mathematics or related fields. How to estimate his or her knowledge? Pay attention to the researches he/she carried out, where findings of these researches were published, and what are his/her contributions. In most cases, for this person coding skills are not required, unless you strive for building a data science team from scratch and DS will form its R&D frame.
Data Engineer Skills
The core skill every data engineer has to possess is a strong knowledge base of software development. They must have a deep understanding of data structures and algorithms. Big data engineer’s skills in coding must be excellent. Especially, pay attention to candidates, who often contribute to open-source projects. If they usually use the same technology stack as you do (Scala, Python etc.), that would be a perfect match.
Data Science Team Manager Skills
This person has to be highly experienced: at least three or five years in managing such teams or working with similar projects to the ones you have. This specialist has to be able to code and has to have a deep technical background. Deep knowledge and practical experience in a code review would be a great advantage. A solid understanding of algorithms is also one of the necessary skills of a good data science team lead.
Skills that are related to all data science team roles depending on a project type:
- Visualisation and reporting;
- Business management;
With all the necessary “must-have” skills determined, we now may proceed to the interview checklist you can utilize whenever you decide to hire a new employee.
How Should You Interview Potential Data Science Team Members
The best DS specialists obviously have deep expertise in one particular domain and at least basic knowledge in one or a couple of other domains. An interview is the most important part of recruitment process, after which you have to make a hiring decision. That is why the best offhand interview is actually a prepared one. So which big data interview questions should you ask and how to determine a professional level of each candidate?
Estimate Their Experience
At this stage, a number of working years is not as important as you may think. What really matters is the list of projects completed by a candidate. Furthermore, the important thing is what results he or she has managed to achieve but not what he/she was doing.
The “What” questions for experience estimation:
What project are you most proud of?
What contribution,attributed to you, positively affected the business on your last position?
What responsibilities will you have working on the position you apply for?
Estimate Their Expertise
To solve problems, the project may face, a candidate has to understand its objectives and what challenges he or she will be dealing with. This understanding is highly important especially for a data science team lead.
The list of big data interview questions for the expertise estimation:
- Could you please describe the case when you have managed to improve existing business processes?
- What was your greatest input for the success of the company you worked for?
- How did you manage to reach this result?
Beyond that, you have to check technical skills of each candidate in advance to avoid situations when a DS professional is not the one you are looking for. There is no doubt that these data science interview tips will help you determine educated specialists, who are real problem-solvers.
How to Measure DS Team Members’ Performance in Action?
The main indicator of all big data teams success is their performance: how fast they can build a specific feature and what quality this feature will have. This feature-based approach will set the specific measuring unit for work estimation. Suppose, a standard timeframe for creating an average feature, let’s say two-four hours, so one DS professional is expected to build 2-4 features a day. This parameter will help you estimate employee’s’ efficiency – how fast they do their work. Deep code review and accurate testing will allow to estimate their effectiveness – how high their performance is.
The key to successful DS team recruiting is the match of your employees’ interests and data science team goals. Their skills have to coincide with the requirements of your project. Furthermore, a well-prepared interview and accurate performance measurement will help you create a well-organized, effective, and efficient team. Remember that commitment, communication, technical skills, experience, and strong expertise are the building blocks of the big data dream team.