About Me

I am currently a Senior Applied Machine Learning Scientist at The Washington Post. My work focuses on revenue optimization-focused projects. I am actively working on using Reinforcement Learning to optimize paywall targeting at The Post. Read more about this project in our Engineering Blog.

I received my Ph.D. in Computer Science from New York University. My dissertation research was in the intersection of Natural Language Processing, Social Media, and Privacy. I was primarily interested in Stylometry (the study of style in text) and how it can be used to mitigate abuse in online communities. View my publications here.

Prior to my Ph.D., I received a M.Sc. in Computer Science from Drexel University. I am originally from Sri Lanka and currently living in the United States.

Skills & Interests

  • Artificial Intelligence
  • Reinforcement Learning
  • Revenue Optimization
  • Natural Language Processing
  • Social Media Analytics
  • Privacy

Resume

Education

2022
New York University
New York

Ph.D. in Computer Science

Thesis: Using Stylometry to Mitigate Abuse in Online Communities

2018
Drexel University
Philadelphia

M.Sc. in Computer Science

2014
University of Moratuwa
Sri Lanka

B.Sc. Engineering

Major: Computer Science

Experience

2022 - Current
The Washington Post

Senior Applied Machine Learning Scientist

Built multiple prediction models focused on revenue optimization and ran randomized control trials and A/B tests. My most significant contribution is the implementation of the Smart Metering algorithm, which currently makes most of the paywalling decisions at The Post. Apart from this, I am involved in designing other key models. I have interviewed, onboarded, and mentored new data scientists.

2015 - 2022
Drexel University & NYU

Research & Teaching Assistantships

I was a research assistant and a teaching assistant at various points during my graduate degrees at Drexel University and New York University. Apart from my dissertation research, I conducted office hours, graded student assignments, and guided student group projects. I conducted lectures on Planning and Plan Recognition and Introduciton to LISP. I also assisted in writing grant proposals together with industry and academic collaborators.

2019 Summer
Facebook

Software Engineer Intern

Working in the Site Integrity - Fake Engagement team, I Implemented a deep learning model (LSTM) to detect suspicious user activity on Facebook. I trained embeddings to represent user actions and worked with several large-scale and real-time database systems (Hive, Presto)

2014-2015
99X Technology
Sri Lanka

Software Engineer

Worked on two commercial products to port legacy web applications to JavaScript (AngularJS, HTML, CSS, C#)

Selected Publications

In the News

Work at The Washington Post

Reinforcement Learning for Paywall Targeting

This blog post explores the reinforcement learning model that I worked on at The Washington Post which predicts whether to display a paywall on an article. The following venues also discuss this work:

  • WaPo Engineering Blog: How the Washington Post Uses Smart Metering to Target Paywalls [Link]
  • The World Association of News Publishers (WAN-IFRA): Data science reading list [Link]
  • International News Media Association (INMA): Diminishing pageviews, a focus on quality journalism indicate a subscription-based future [Link]
  • Svet Petkov: Paywall and SEO: Types and Implementation Guide [Link]

Press coverage of Dissertation Research

Instagram Pods:

This coverage is based on our paper titled "The pod people: Understanding manipulation of social media popularity via reciprocity abuse", in which we study an underground ecosystem, mostly operating off of Telegram, that uses reciprocity abuse to manipulate the popularity of Instagram content.

  • TechCrunch (2020): Instagram 'pods' game the algorithm by coordinating likes and comments on millions of posts [Link]
  • NYU Engineering (2020): Researchers use machine learning to unearth underground Instagram 'pods' [Link]
  • Sci-n-Tech (2020): Underground Pods: Instagram account manipulators [Link]
  • Times Now News (2020): How Instagram 'pods' manipulate algorithms and artificially boost content popularity [Link]
  • MediaPost (2020): 'Pod' People Are Gaming Instagram Engagement Rates [Link]
  • Ombre Digital (2020): Pod People Are Gaming Instagram Engagement Rates [Link]
  • Builtin (2020): Instagram Bots Are Frowned Upon — but They Work [Link]
  • Yahoo Noticias (2020): Descubren 400 "rebaños" en Instagram, grupos de cientos de cuentas que engañan al algoritmo inflando likes y comentarios [Link]