Data Science Team

Five Ways to Build a Data Science Team for your Organization

As a data science consultant for years, I would like to share my viewpoints and experiences of developing the capability for a data science team for any organization.

(NOTE: this article is aimed to share building data science for general business.  However, it is not for helping you to build a Data Science consultancy or a start-up focused on AI.  These companies should take other methods to build their technical team.)

Basically, there are 5 different approaches to have the Data Science resources:

  1. Internal Staff Transformation – deliver training to internal staff in order to build their data analytic skills
  2. Employing New Staff with Data Science Experience – hire new people with Data Science background
  3. Hiring External Data Science Consultant – engage your Data Science projects with external parties like freelancers or consulting companies.
  4. Data Science as a Service (DSaaS) – purchase services from vendor such as Snowflake
  5. Crowdsourcing in the Internet – host a competition for solving a Data Science problem like doing it on Kaggle

Which is the best for you / your organization?  Again, it depends…

Let’s compare the PROs and CONs across different approach.

Buy Me A Coffee

Internal Staff Transformation

This approach is fit for industries requiring deep domain knowledge such as Medical Care, Genetics, etc.  Established companies are seeking to introduce Data Analytics and Data Science to support their development.

Pros

  • Strong Domain Knowledge
  • Knowledge of Business Process New
  • New Talent Raises the Team Performance immediately
  • Gradually increase the quality of services

Cons

  • Risk of homogenous thinking
  • May struggle the quality of service if the team could not pick up relevant technical skills
  • Some Team Members may resist changes

Unfortunately, this approach is not popular in Asia due to employers are afraid the investment of training will be lost if the staff going to somewhere else or even their competitors.

 

Employing New Staff with Data Science Experience

The first major reason to take this approach is a start-up company where data is a product.  It is also suitable for company to start new data science projects focused on advanced analytics.  Deep domain knowledge is less critical than technical know-how.

Pros

  • Control Over Skill-sets
  • More Flexibility
  • High Quality of Service
  • Lower Risk Compared to External Parties / Services

Cons

  • Hiring and Knowledge Transfer are time-consuming
  • Time required to find and hire right team members
  • Managing Skills on New Talents

This is taking time for new team members to learn the domain knowledge and familiar with the working environment.  If there is no Data Science experience before, it is difficult for managers to handle and manage the new staff members.

 

Hiring External Data Science Consultant

First of all, there are lots of different ways to hire external resources.  It is possible to hire independent consultants, academic experts, or freelancers to support your Data Science needs.

Pros

  • Highly Skilled People in-place
  • More Flexibility with Part-time Resources
  • Helping to build from-scratch or recruit a lead Data Scientist / a Data Science team
  • High Quality of Service
  • Quick Answer for Specific Questions
  • Introduce New Technologies / Knowledge Immediately

Cons

  • High Risks
  • Difficult to Manage
  • Internal Resistance from Existing Staff
  • Time required to find and hire right consultants

It is not easy to find skillful freelancers and different risks including the commitment of the external parties.  For this method, it is better for the managers with strong Data Science knowledge or managers are Data Scientists.  So, it is more likely to solve short-term team members shortage rather than increasing the capabilities in the long-run.

 

Data Science as a Service (DSaaS)

If an organization is not willing to change any existing structure, it is suitable to buy a solution directly from a vendor.  It is just like purchasing a software platform without customization.  It is vital to consider the SLAs from the vendor able to meet the business requirements.

Pros

  • Able to scale on demand
  • May get better service levels than in-house
  • Learn from outside experts

Cons

  • Provider may not understand company’s unique processes
  • Difficult to bring expertise back in-house
  • Decreasing quality of service over time

 

Crowdsourcing

Crowdsourcing a solution is only possible for a very specific problem.  The business is willing to listen to different opinions from different people.  Also, the problem is able to be shared to the general public without serious risk for the business on the commercial confidentiality.  There should always be a “Plan B” available if there is no “acceptable” solution in the market.

Pros

  • Leverage Wisdom of the crowds
  • Diverse perspectives
  • Lower Cost
  • Fast Results

Cons

  • No SLA; value not guaranteed
  • Difficult to design the “Open” problem
  • Difficult for Domain Intensive Tasks
  • Crowd failure may happen (adds cost)

 

Conclusion

Personally, any one of the above approaches alone may NOT be your best-fit solution.  Also, it is possible to have a mixture to employer different approaches in order to have a stronger team.

If you would like to seek for my advice, it is welcome to reach me via the contact form.

 

Samuel Sum

Data Scientist / Chartered Management Consultant (AS, CDS, SDi)

0Shares