How to build an effective data science team for a tech startup

null

Within the technology sector, numerous organisations find it difficult to deliver effective and disciplined data science programs. Despite providing a promising solution, many pilots often fail to evolve into real, actionable results. A successful data science program requires four key components: skills, data, technology and direction. 

As well as this, startups need to bring on board experienced staff and balance this with junior hires, whilst also creating the perfect environment for your team to thrive. A startup needs to maintain long-term vision, even under the pressure of short-term needs. It’s therefore vital to adhere to your plans and ensure the talent you hire is also aligned with this strategy.

Source diverse skill sets

Trying to build an effective data science team with only data scientists is like trying to play a football game with just strikers— if your team is to be successful, a variety of skill sets are required. Firstly, you need data scientists with an understanding of your domain and technology, as well as a thorough knowledge of data analytics and machine learning. Good data scientists also require strong communications skills, which means that a combination of these qualities comprises a rare skill set. They tend to be hybrids and are highly flexible, so it’s important to ensure they aren’t used in the wrong role.

Whilst your data scientists create the solution, you need data engineers to ensure the technology runs when you need it. These hires require excellent development skills and data knowledge as they source data and produce the solution, as well as covering the scaling of complex custom algorithms.

An effective infrastructure needs someone to put it in place and make timely decisions, dependent upon data. For this you need to bring data architects on board. The role demands a detailed understanding of your solutions, making these team members key to providing the necessary interfaces to the operations systems and the security which supports the project.

Since data science solutions are often highly abstract, you need business analysts to turn their minds to the matter. This job requires good communication skills to ensure your end-user has a sufficient understanding of the solution. The most effective business analysts usually have a technical background as they must fully understand the functionality of the data science methods, even though they don’t need to know the nuances of how they are deployed.

Balance your team experience

As well as a solid combination of skills, your team should also display a balance of experience. When hiring junior staff, don’t simply choose based on technical aptitude, but consider candidates who have the ability to think through a problem. Technical skills can be taught to the right staff in weeks, whereas it’s much more difficult to teach logic and abstract thinking.

Make sure you’re looking further afield than just data science degree graduates – they may know about the technology you use but that doesn’t mean they understand how to apply it. Some of your best candidates may come from a background in subjects like Astrophysics, which requires the skill to look at a problem as a whole and describe it in a series of logical statements – proficiencies that are essential to becoming a data scientist.

Build trust in your team

Once you’ve found the right skill sets, you need to set up an environment that allows your data science team to work effectively. The first thing is to establish trust and security because good data science requires access to high volumes of real data. For many issues, anonymising data for security purposes reduces the quality of what you can find. When necessary, explain to your clients that whilst working with anonymised data is doable, it can result in lower quality solutions and increased cost development. With the right balance between effective security and efficient solutions, you will continue to build the trust of your clients.

Provide support and direction

Always invest in training courses, as the skills your staff learn in these will pay off for many years. If a general understanding is required then external courses may be appropriate, but for specialised skill development, you should consider building your own training courses. It’s important to take a collaborative approach to this, encouraging all your trainees to contribute to your course to ensure that no area is missed.

Mentoring is also key and provides your staff with the direction they may need to fulfil their aspirations, which helps to keep your staff motivated and determined. One of the most effective ways to expand someone’s knowledge is to give them exposure to those with maximum experience, so place junior and senior staff together on every project.

Problem-solving

In order to grow as a startup, you need to ensure that your business is constantly adding value. The best way to do this is to maintain a focus on the right mechanisms that will help to drive innovation. Regular meetings dedicated to this can be useful, where staff members can present on new ideas and their use for the future, as well as ways that new research or new data can be utilised differently. Insights are not simply gained through established stages, but must often be developed through revisiting and reassessing earlier information, and reviewing new feedback. This means teams can be more agile along the path to achieving goals and can hone problem-solving skills.

Structure

While the organisation of a data science team varies based on the size and setup of the company, common factors should be considered. A centralised or shared model, where scientists are in the same location and work together, serves to enhance collaboration and the cohesiveness of the team so that ideas can be shared and developed, and problems can promptly be addressed. When teams and companies grow too big for this to be maintained, you should ensure you are spreading business units so that managers, scientists and stakeholders can all still interact when needed. You also need to ensure you invest in effective business communication tools so that teams can communicate well across geographical distances and time zones.

In summary, a successful data science team is created from a collaboration of skills and experience. Once you’ve found the right staff, be sure to provide them with training and development opportunities to continually grow. Combining an effective data science team with your technology will support your startup on its journey to success.

Felix Hoddinott, founding member, Quantexa
Image Credit: Geralt / Pixabay