Jul 12th 2021
Meet James and the Data Science Department
Title: Engineering Manager, Machine Learning
Department: Data Science
Introducing the Data Science department
The data science team applies predictive modeling to tackle a range of business problems at Vungle. We take models and theories from research and translate them into production systems that decision at scale.
Why is the Data Science team vital to Vungle?
The company operates in fiercely competitive ad auction markets and relies on data to drive how it values, bids for, ranks, and serves ad impressions. We try to encapsulate business intuition into machine learning models that are able to scale to billions of samples and millions of feature interactions.
What are some initiatives the Data Science team is working on right now?
Our recommender system drives which ad is chosen for a particular user and is at the heart of Vungle’s ad serving pipeline. It incorporates conversion rate modeling and user LTV modeling and has to contend with challenges around data sparsity, high dimensionality, reinforcement learning, Bayesian inference, and online learning.
What makes someone successful on this team?
Curiosity and a can-do attitude are crucial for growth in the team. There’s so much to learn across the fields of ML, statistics, and engineering, and the fields are constantly changing, so you have to love math, enjoy reading research, and constantly learn outside of school. But you have to be practical too. We do a lot of our own implementations, taking whatever tools, wiring them together, and debugging them to clean, transform, and fit our data. You have to be willing to come out of the ivory tower and get approximate solutions working for business deadlines.
Can you share a fun fact about you or the team?
I’ve once held my breath for four-and-a-half minutes freediving in a former life in Singapore. It’s come in handy recently for using public restrooms during COVID-19.