Assembling software development freelancer comprised teams is a challenging task for a project manager. In an Agile environment, this is even more challenging because of the cross-functional character and multi-disciplinary requirements such methodology enforces. The project manager needs to identify individuals that are highly compatible from a collaborative point of view and have the skills to match project complexity. This is a lengthy endeavor which is prone to human bias as the collaborative nature of individuals that have never worked together is difficult to evaluate. The selection process also has considerable restrictions imposed by the sheer volume of information a human decision maker can process and analyze. Computerized data analysis, on the other hand, is free of bias and it can process large data sets very fast. For the current research process data was collected automatically from a freelance platform that allows free access to developers’ profiles. The targeted platforms make information available about freelancer’s skills, experience, background, and education. Data was validated, ensuring it’s correct and relevant. Data was normalized to ensure it is easily quantifiable and can be measured using a common scale. Statistical methods were used to predict the data that was missing or incomplete. Based on the identified criteria a template was built to match the profile of the ideal team member. Machine learning techniques were used to group team members that would work together effectively from a collaborative point of view. An indicator was built to benchmark the team member’s profile against the template and determine the highest ranking candidates. Conclusions were formulated regarding the topic of using profiling to assemble collaborative agile freelance software development teams. Research process limitations were enunciated and future research topics were submitted for debate.
