Bio

I lead the Account and Behavioral Abuse team at YouTube Trust & Safety.

At Account and Behavioral Abuse we are working on YouTube’s (and the industry’s!) #1 priority – defining, detecting & preventing scaled abuse on our platform. It’s hard, it’s interesting, and it matters. We work across verticals like spam, adult, misinformation, scam, election influence etc. We focus on understanding incentives & behaviors to create strategies to combat bad actors, while enabling YouTube to be a place safe for users, creators and advertisers to thrive.

Prior to that I was a research manager at the AI Integrity team, at Facebook AI. Our work there spanned a broad range of applications: hate(ful) speech, misinformation, detecting political ads (for ensuring transparency reports on them), sales of regulated goods, as well as very high severity and low prevalence problems like CSAM and suicide prevention.

Before Facebook, I used to work on machine learning and data mining techniques and applied them to Automatic Speech Recognition and Spoken Language Understanding at Microsoft. Massively parallel data processing, large scale text (or query log) mining for statistical semantics, personalization and contextualization of language models, and things like that.

Even earlier, I used to work on search: part of of the Bing whole page relevance team, and as a scientist at Yahoo! Labs (2008-2011).

co-authored more than 50 papers, and co-inventor of 14 US patents. Reviewer for many journals and conferences and former Associate Editor of IEEE Transactions on Neural Networks.