Honey

Honey / PayPal · 5 yr 1 mo

Jul 2021 – May 2022

Designed ML system to optimize the allocation and pricing of "Offers", Honey's ads product, improving monetization and enabling us to grow offer volume an order of magnitude
Built the shared ML feature store system serving hundreds of terabytes of versioned text embeddings, image representations, and other numerical features to online ML-powered products across the company

Sep 2018 – Jun 2021

Invented novel distributed hybrid fulltext + semantic search engine for next gen of Honey's product search engine before industry solutions existed
Built personalized product recommendation engine serving relevant product suggestions to 30M users despite extremely sparse interaction data
Built "Similar Products" service finding related and near-identical products for every product in the catalog
Solved fine-grained product categorization of 100M's diverse products from 10K's stores into a unified taxonomy of ~25k leaf categories w/ high accuracy

Jan 2018 – Sep 2018

Designed and built Honey's first production ready universal product search engine enabling accurate semantic search of 100M's of rapidly changing products from thousands of stores
Built tooling for annotation and curation of golden dataset to measure search accuracy across algo improvements

May 2017 – Dec 2017

Used Dataflow, BigQuery, Python, SKLearn to analyze trillions of events, derive important insights, and inform product development decisions. Presented key findings to senior leadership in weekly data meeting