MS in Data Science - Seminar Series
Andrew Chamberlain: Predicting Salaries with Deep Neural Networks at Glassdoor
Introduction of the speaker:Andrew Chamberlain is Sr. Director of Product for Machine Learning at Glassdoor, the online jobs platform. He is an applied economist specializing in the data science of online review platforms. At Glassdoor, he is responsible for crafting the vision, strategy and product development for machine learning, algorithms, and personalization across the company’s full suite of consumer- and employer-facing products. In addition, he founded the company’s Economic Research group and served as Chief Economist for more than 6 years.
Andrew received his Ph.D. in Economics from the University of California, San Diego in 2014, where he received four consecutive teaching awards for undergraduate instruction. A Seattle native, Andrew earned his B.A. degrees in Economics and Business Administration from the University of Washington, Seattle in 2001.
Abstract:This talk will walk through Glassdoor's machine learning approach to modeling predicted salaries for millions of job postings and the production deep neural net (DNN) architecture that powers our salary products (on pages like this). We'll discuss how we use text embeddings, key statistical concepts underpinning our model, and model guardrails we've developed to ensure good front-end user experiences in our salary products.