pooja sethi

hello there
I'm Pooja. I'm a Software Engineer at Facebook and a @UWCSE '17 graduate. I'm from sunny Seattle, where we have coffee and code for breakfast.

About Me

I recently graduated -- thanks to all the badass and brilliant people who constantly inspire me. Now, I'm working on fun NLP problems as a software engineer at Facebook in the Bay Area. I'm still learning the whole "adulting" thing... at least I can say I have avocado toast down.

The best way to reach me is via email.

A pdf of my resume is available here.

Industry Experience


Software Engineer · October 2017 - Present · Menlo Park, CA
  • Facebook Messenger Assistant (M) Team.
  • ML model development and data engineering for Natural Language Understanding (NLU).
Natural Language Processing PyTorch Python


Software Engineer Intern · September 2016 - December 2016 · Menlo Park, CA
  • Instagram Explore team. Machine learning to suggest personalized media to you.
  • Created Hive data pipeline, trained a gradient boosted decision tree CTR model for video channel cover media on Explore, and evaluated model with A/B test. Increased immersive viewer time spent by 8%.
  • Developed new features for ranking Explore Stories.
Python Machine Learning Hive


Software Engineer Intern · June 2016 - September 2016 · Mountain View, CA
  • Payments Accounting team. The engineering behind Google Finance.
  • Built an internal tool to visualize accounting impact (credits and debits) of events such as payments, invoices, etc. as a transaction is processed.
  • Integrating tool with Google's F1 database.
Java AngularJS


Software Engineer Intern · January 2016 - March 2016 · Bellevue, WA
  • Bing Ranking Service team. Computes rank of web pages for optimally ordering Bing search results.
  • Built a dynamically linked library (DLL) to support rapid feature development for ranking machine learning models, such as decision trees and neural networks.
  • Allowed quick turnaround of features, reducing the iteration time for trial in production beds from several weeks to several days.
C++ Machine Learning


Software Engineer Intern in Tools and Infrastrucutre · September 2015 - December 2015 · Kirkland, WA
  • Spanner Tools and Infrastructure team. Develops tools supporting Spanner, Google's globally distributed database.
  • Designed and developed a graph-based model for GoogleSQL random schema generation to complete Spanner's end-to-end integration testing framework. Complements random query random generator used to rapidly identify bugs in Spanner.
  • Created an API that allows reuse of schema generator across multiple Google storage platforms such as Dremel and F1.
C++ Databases


Software Engineer Intern · June 2015 - September 2015 · Menlo Park, CA
  • Messenger Payments team.
  • Built P2P group sends for Android, which allows you to send money to friends in a group conversation. Built frontend and backend. Coordinated launch with iOS and www teams.
  • Built framework for native Android risk verification flow, migrating away from old mobile site.
Android Java Mercurial

Research Experience

University of Washington CSE NLP xLab

April 2017 - September 2017 (Part-time) · Seattle, WA
  • Started and led Conversational Question Generation (CQG) project i.e. learning to ask questions for chit chat.
  • Developed models for question-asking, via adaptation of the encoder-decoder architecture.
  • Project was selected for Facebook ParlAI research award grant.
  • Advised by Yejin Choi.
Natural Language Processing PyTorch Python

Allen Institute for Artificial Intelligence (AI2)

April 2016 - June 2016 (Part-time) · Seattle, WA
  • Developing a model for learning perturbations of fourth grade level science questions. A perturbation is a slight modification of a questions that drills into whether a student actually understood the concept being tested or not.
  • Advised by Been Kim.
Python Machine Learning Data Science

University of Washington CSE ICTD Lab

April 2014 - June 2017 · Seattle, WA
Android Java


  • (To Appear in ACM CHI 2018) Aditya Vashistha, Pooja Sethi, and Richard Anderson. 2018. Bspeak: An Accessible Voice-based Crowdsourcing Marketplace for Low-Income Blind People.
  • ACM CHI 2017 Best Paper Honorable Mention Aditya Vashistha, Pooja Sethi, and Richard Anderson. 2017. Respeak: A Voice-based, Crowd-powered Speech Transcription System.
  • Awards

    • Outstanding Honors Undergraduate Thesis Award
      Awarded for best undergraduate thesis in UW CSE department.
    • Outstanding Computer Engineering Senior Award
      Awarded to top two UW CE seniors.
    • Society of Women Engineers Outstanding Female Engineer Award
      Awarded yearly to one undergrad and grad woman in the UW CSE department.
    • Facebook F8 Scholarship
    • TUNE House Scholarship
    • Qualified for Tau Beta Pi & IEEE HKN
    • NASA Space Grant Scholarship
    • Mary Gates Scholar
    • University of Washington Honors Program Scholarship
    • Martin Family Foundation Honors Scholarship
    • Granite Falls High School Valedictorian (GPA 4.0/4.0)



    • Vice Chair for women's student chapter of ACM.
    • Organized activities that promote the equal representation of women in computing.

    Read Copy Update

    • Started the first undergrad computer science paper reading group at UW CSE.
    • Led a weekly discussion group covering papers on distributed systems, machine learning, and programming languages.


    • Internal Relations officer for UW CSE ACM.
    • Planned events for CSE undergrads and was the main liason between faculty + advisors and students.

    CSE 14x TA

    • Taught section of ~20 students for CSE 143, UW's second-level intro to programming/Java class.
    • Topics covered include recursion, linked lists, stacks/queues, binary trees, basic complexity, etc.


    As a kid, every Sunday I would drink chai with my mom and dad and read the Sunday comics. Now I mostly read xkcd (I swear, there's a relevant xkcd for every situation in my life) and a lot of books. I still drink a lot of chai.

    I love long distance running and have also recently been learning how to climb.

    My name is worth 22 points in Scrabble.

    My Meyers-Briggs personality type is INTJ.

    © Pooja Sethi