Curriculum Vitae

You can download my résumé from the pdf button

General Information

Full Name Sabyasachee Baruah
Date of Birth 13th January 1994
Place of Birth Guwahati, India
Languages English, Hindi, Assamese

Education

  • 2018-now

    Los Angeles, USA
    PhD
    University of Southern California
    Computer Science
    • Advisor - Dr. Shrikanth Narayanan
    • Signal Analysis & Interpretation Laboratory
    • Expected Graduation - Dec 2024 (Fall)
    • GPA - 3.95/4
  • 2012-2018

    Kharagpur, India
    Bachelors & Masters
    Indian Institute of Technology Kharagpur
    Computer Science & Engineering
    • Secured 10/10 GPA in second semester
    • CGPA - 9.38/10
  • 2000-2012

    Guwahati, India
    School
    Maharishi Vidya Mandir
    • Grade 1 to 12

Work Experience

  • May-Aug 2022

    Los Angeles, USA
    Student Researcher
    Google
    Perception Team
    • Worked on multimodal coreference in English news videos
    • Designed an annotation interface to label person names of face tracks
  • May-July 2016

    Los Angeles, USA
    Research Intern
    University of Southern California
    • Developed a regression model for continuous valence and arousal prediction in movie clips
    • Trained a model on discrete-emotion video clips and used it as the base model for gradient boosting
  • May-July 2015

    Bengaluru, India
    Research Intern
    Adobe Big Data Experience Lab
    • Quantified sufficiency, relevance, and coverage of time-series data
    • Designed a ranking and recommendation system of data tables for journalists

Projects

  • 2023
    Character Attribute Extraction
    • Explored different prompting methods on large language models to extract attribute frames of characters from movie screenplays
    • Published work at ICASSP 2024
  • 2022
    Multimodal Coreference Resolution
    • Developed an annotation interface to label face tracks with coreferring person names
    • Tested on English news videos of YouTube-8M
    • Work done as part of Student Research internship at Google
  • 2020-2023
    Character Coreference Resolution
    • Curated a coreference-annotated dataset of movie screenplays
    • Developed scalable inference approaches to adapt coreference resolution to long documents
    • Published work at ACL Findings (2020 & 2023)
  • 2022
    Media Representation of Professions
    • Developed a searchable taxonomy of professions for media content
    • Analyzed the frequency and sentiment trends of professions over 7 decades and 136K movies and TV-shows
    • Published work at PLOS ONE (2022) journal