End-to-End Data Science to Drive Business Decisions at LinkedIn with Burcu Baran - TWiML Talk #256

Team Coordination

LinkedIn's data science teams employ a unique coordination strategy to tackle machine learning challenges. Burcu Baran explains that her horizontal team collaborates with vertical teams, like LinkedIn Talent, to convert business problems into mathematical models. This approach ensures that all teams are in constant communication, allowing them to address complex issues effectively 1. She notes, "When their data science team has a machine learning problem that's more complicated than expected, they bring it to our team, and we solve it together" 1. This collaborative framework helps prevent missteps in problem formulation and enhances the overall problem-solving process.

   

Communication

Effective communication is crucial for problem-solving within LinkedIn's data science teams. Burcu Baran highlights that frequent interactions between teams help identify and rectify issues early in the modeling process 1. She describes a systematic debugging process that involves both manual checks and platform-based solutions to ensure accuracy in model development 2. Baran emphasizes, "We are very careful about the problem. For example, on B2B business, or even B2C, like acquiring a new customer or empowering a new customer, we have done those models before, so we know how to label logic" 2. This meticulous approach minimizes errors and enhances model reliability.