Optimizing Recommendation Systems (feat. Etsy) | Online

About this event

Talk Abstracts:

Data-Science'ing on the Cheap with Alex Kaos, Data Scientist at Dataiku: Whether you are a data science student or a startup founder, you will have a limited budget to do data science with and an enormous amount of tools at your disposal. This talk explores the hardware that all DS algorithms end up running on and offers tips and tricks to maximize what you can do with your money. When does it make sense to use the cloud? Can you do it without going broke? Let's find out.

Optimizing Recommendation Systems at Etsy with Moumita Bhattacharya, Sr. Data Scientist at Etsy: Traditionally, recommender systems for e-commerce platforms are designed to optimize for relevance (e.g., purchase or click probability). Although such recommendations typically align with users’ interests, they may not necessarily generate the highest profit for the platform. In this talk, I will present a novel revenue model that we developed here at Etsy, which jointly optimizes both for probability of purchase and profit. The model is tested on a recommendation module at Etsy. We show that the proposed model outperforms several baselines by increasing offline metrics associated with both relevance and profit. In addition to the above approach, this talk will also provide an overview of the recommendation systems at Etsy and our journey from linear ranking models to a non-linear deep neural network ranking model.

Speaker bio: Moumita Bhattacharya is a Senior Data Scientist at Etsy, a two-sided marketplace for buyers and sellers. At Etsy, Moumita develops recommendation systems to show relevant items to Etsy users. Recently, she developed a ranking method to improve conversion rates and gross merchandise sales of the company. As a part of another project, she worked on a new module recommending shops for a user on the Etsy app, which led to increased conversion rates. She is currently building custom objective functions to optimize for metrics beyond relevance and is also incorporating different contexts in recommendations. Moumita has a PhD in Computer Science with a focus on Machine Learning and its applications in disease prediction and patient risk stratification. During her PhD, she developed machine learning methods for risk prediction of heart disease and chronic kidney disease, in collaboration with cardiologists from John Hopkins and nephrologists from the largest hospital in Delaware. Prior to grad school, Moumita worked as a software engineer for three years in the financial services industry.

Alex Kaos is a Data & AI systems architect at Dataiku where he helps deploy DSS-powered AI solutions tailored to small and large companies as well as train IT administrators to run it in optimal conditions, write plugins, etc. Prior to his time at Dataiku, he worked as a Data and systems engineer in the telecom, pharma and construction industries. Alex got his start tuning hardware building gaming computers with limited budgets 25 years ago and got bit by the optimization bug. He enjoys spending days researching and spec’ing goldilocks computer systems for every purpose.

Please note: In order to join the event, you MUST RSVP on both Meetup and the General Assembly website here: TBD

Tentative Schedule: 6:30pm: Pizza + Beer networking 7:00pm: Data-Science'ing on the Cheap with Alex Kaos, Data Scientist at Dataiku 7:30pm: Optimizing Recommendation Systems at Etsy with Moumita Bhattacharya, Sr. Data Scientist at Etsy