DEPLOYING PROCESS AUTOMATION FOR MEETUP GROUPS APPROVAL

For this meetup, we're going a little meta with Meetup's own Shayak Banerjee, a senior Machine Learning Engineer at the event platform discussing process automation. Tentative Schedule: 6:30pm: Pizza + Beer networking 7:00pm: TBD with Data Scientist at Dataiku 7:30pm: Deploying Review Process Automation for Meetup Groups Abstracts: The Meetup platform sees in excess of 500 groups created daily across the world. A team of humans review each group to ensure they adhere to community guidelines, and disapproves groups which do not. In this talk we’ll discuss an internal machine learning tool called Fastpass that auto-approves a small fraction of these groups based on mimicking the understanding of a human reviewer. We will talk about the model itself (features + XGBoost), the nuances and pitfalls of determining “approvability”, and how we deployed and operationalized it using Spark, Airflow and DynamoDB. We’ll touch on some of the precision/recall tradeoffs driven by product choices, and we’ll end with a look at how we set up continuous monitoring on the model performance.