About
I am a computer scientist and social computing researcher with a passion for understanding and mitigating the societal impacts of AI, particularly in the context of labor and technology policy. My research spans across societal impacts of AI, explainable AI, and multimodal machine learning.
Bio
Varun Rao, is a Ph.D. student in Computer Science at Princeton University advised by Prof. Andrés Monroy-Hernández, currently part of the Workers Algorithm Observatory (WAO). His research combines system building and empirical studies to understand and mitigate AI's impact on labor. Current projects include FairFare, examining black-box rideshare algorithms, and OpenDeli, exploring decentralized food delivery platforms. Earlier in his Ph.D., Varun's work exposed discrimination in job ad image selection on platforms like Facebook, revealing how this practice circumvents existing protections.
In his professional experience, Varun worked as an Applied Scientist at Amazon, developing explainable vision-language models for customer protection and seller trust. During his internship at Apple, he enhanced OCR solutions for handwritten text recognition
Varun engages in tech policy outreach, influencing legislation on rideshare transparency. He is passionate about CS Education Research focusing on introductory algorithms courses and developing concept inventories. His research has earned several recognitions, including the COLING 2020 Outstanding Paper Award, MIT/DARPA Graph Challenge Honorable Mention 2017, and the Hamsa Kartik Alumni Award for ranking first in his undergraduate class.
Affiliations at Princeton: HCI Group, Center for Information Technology Policy (CITP), Computer Science
News
-
[Oct
2024]New paper accepted to SIGCSE 2025: Construction and Preliminary Validation of a Dynamic Programming Concept Inventory
Wonderful collaboration with co-authors from UC Irvine, Dickinson College, USC, Northeastern University, Utah State University, Boston University
-
[September
2024]- New paper: Rideshare Transparency: Translating Gig Worker Insights on AI Platform Design to Policy (to appear in CSCW 2025). [preprint]
TLDR: We use a novel methodology of LLM-Driven Reddit Data Analysis and Interviews to Gather Gig Worker Insights on AI Platform Design and translate to a policy solution, proposing the first rideshare transparency report, similar to those for social media platforms and AI models. - Invited Talk at National Employment Law Project (NELP) on "Making Rideshare Governable". [slides]
- I won a Graduate Fellowship of Social Data Science at Data Driven Social Science Initiative, Princeton University.
- Submitted a response to EU Multi-stakeholder Consultation "Future-Proof AI Act: Trustworthy General-Purpose AI" - website, response.
- New paper: Rideshare Transparency: Translating Gig Worker Insights on AI Platform Design to Policy (to appear in CSCW 2025). [preprint]
-
[July
2024]I am co-organizing a workshop on Worker Data Collectives at ACM CSCW 2024, Costa Rica.
Apply here -
[June
2024]New Preprint, Blogpost and Policy Memo calling for Rideshare Transparency.
Grateful to have worked with the CITP Tech Policy Clinic, Colorado Fiscal Institute and Towards Justice and helped influence bill language of Senate Bill 24-75 Transportation Network Company Transparency
From Feb 2025, rideshare platforms in Colorado must disclose rider prices and rider wages at dropoff. A big transparency win!
-
[May
2024]New preprint on using LLMs to translate unstructured online forum discussions to structured text resembling a survey.
See QuaLLM.
Presented this at the CHI LLM as Research Tools Workshop at CHI 2024
-
[May
2024]I passed the Princeton Computer Science PhD General Exam!
Committee: Andrés Monroy-Hernández, Peter Henderson, Arvind Narayanan
Title: Societal Impacts of AI on Labor
Presentation Assets: Slides, Reading List, Notes
Education
Princeton University
Doctor of Philosophy in Computer Science, Aug 2022 - Present
Research: Societal Impacts of AI on Labor
Selected Coursework: HCI (TA), Technology Policy and Law, Privacy, Ethics of Computing (TA), Responsible AI (TA)
Carnegie Mellon University
Master of Science in Electrical and Computer Engineering, Dec 2019
GPA: 3.96/4.0
Coursework: Computer Vision, Machine Learning (PhD), Convex Optimization, Security and Fairness in Deep Learning, Advanced Multimodal Machine Learning, Foundations of Cloud and ML Infrastructure
Ongoing Projects
Workers Algorithm Observatory (WAO)
Understanding gig worker concerns and building tools to measure and mitigate AI and algorithmic systems harms; Focus on rideshare and delivery platforms. Contributed to FairFare Project.
Academic paper in progress
Project WebsiteOpenDeli
Designing a decentralized food delivery protocol and a reference implementation consisting of a frontend mobile app, backend and admin interface.
Academic paper in progress
Project WebsitePublications
Societal Impacts of AI and Technology
-
FairFare: A Tool for Crowdsourcing Rideshare Worker Data to Empower Labor Organizers
Under review (CHI 2025) -
Rideshare Transparency: Translating Gig Worker Insights on AI Platform Design to Policy
CSCW 2025 | Link -
Worker Data Collectives as a means to Improve Accountability, Combat Surveillance and Reduce Inequalities
CSCW 2024 Workshops | Link -
QuaLLM: An LLM-based Framework to Extract Quantitative Insights from Online Forums
CHI 2024 LLMs as Research Tools Workshop | Link -
Discrimination through Image Selection by Job Advertisers on Facebook
ACM Conference on Fairness, Accountability, and Transparency (FAccT'23) | Link | Princeton SPIA Press -
I would have to evaluate their objections: Privacy tensions between smart home device owners and incidental users
Privacy Enhancing Technologies Symposium (PETS 2021) | Link
Explainable AI and Multimodal Machine Learning
-
RAVEN: Multitask Retrieval Augmented Vision-Language Learning
In submission | Link -
A First Look: Towards Explainable TextVQA Models via Visual and Textual Explanations
Workshop on Multimodal Artificial Intelligence (MAI), NAACL 2021 | Link -
Misspelling Detection from Noisy Product Images
Conference on Computational Linguistics - Industry Track (COLING 2020) - Outstanding Paper Award | Link -
Gamification of a Visual Question Answer System
Technology for Education (T4E) IEEE 2018, and VQA Challenge and Visual Dialog Workshop, CVPR 2018 | Link -
Extracting and Visualizing Character Associations in Literary Fiction using Association Rule Learning
Advances in Computing, Communications and Informatics (ICACCI) IEEE 2018 | Link
High Performance Computing
-
First look: Linear algebra-based triangle counting without matrix multiplication
High Performance Extreme Computing Conference (HPEC), IEEE 2017 | Link
Computer Science Education
-
Construction and Preliminary Validation of a Dynamic Programming Concept Inventory
Technical Symposium on Computer Science Education (SIGCSE 2025) -
What is an Algorithms Course? Survey Results of Introductory Undergraduate Algorithms Courses in the U.S.
Technical Symposium on Computer Science Education (SIGCSE 2023) | Link
Teaching Experience
-
COS436 Human Computer Interaction
Professor: Andrés Monroy-Hernández
Princeton University, Fall 2024
Course Website -
COS598I Responsible AI in Societal Deployment
Professor: Lydia Liu
Princeton University, Spring 2024
Course Website -
COS350 Ethics of Computing
Professor: Arvind Narayanan
Princeton University, Fall 2023
Course Website
Professional Experience
Applied Scientist II Intern
Amazon Web Services, AWS Bedrock
May - Aug 2023
- Showed that captioning, VQA and image classification can be modeled through a unified retrieval augmented encoder-decoder vision-language architecture, with no pretraining and additional trainable parameters; paper under review.
- Experiments demonstrate the benefits on image captioning (+1 CIDEr on MSCOCO, +4 CIDEr on NoCaps) and VQA (+3 % accuracy on VQA v2) tasks compared to the non-retrieval baselines.
Applied Scientist II
Amazon.com Inc, Multimodal AI - Product Assurance, Risk and Security (PARS)
Feb 2020 - July 2021
- Explainable multimodal text-in-image and classification models to help protect customers and build seller trust.
- Invited Talk "Explainable Multimodal TextVQA Models" - Amazon Machine Learning Conference (AMLC) 2020
- Outstanding Paper Award for "Misspelling Detection from Noisy Product Images" (top 2.5% - 16/644) at COLING 2020
Computer Vision Intern
Apple, Inc - Core Recognition Team
May - Aug 2019
- Enhanced the accuracy of the OCR system by up to 28% and expanded support for handwritten text.
Software Development Intern
Akamai Technologies - Analytics Database Team
Feb - Apr 2018
- Designed and implemented a release checklisting framework that reduced the time for checklisting in a sprint by up to 2 days.
Awards
-
Graduate Fellow of Social Data Science 2024
Awarded by the Data Driven Social Science Initiative, Princeton University
-
COLING 2020 Outstanding Paper Award
For the paper "Misspelling Detection from Noisy Product Images"
-
MIT/DARPA Graph Challenge 2017 Honorable Mention
-
Gold Medalist and Hamsa Kartik Alumni Award
For Rank 1 in CSE Department, PES University 2018
Broader Outreach
-
Lead a response to EU Multi-stakeholder Consultation "Future-Proof AI Act: Trustworthy General-Purpose AI"
Discussions on Reddit different industries e.g. creatives, professionals, educators, which shows AI’s effect on job is more nuanced than simple job displacement, with evidence of workers adapting to AI-driven changes by leveraging AI tools or transitioning to new roles
Website | Response. -
Influenced bill language of U.S. first rideshare transparency law
Passed in Colorado, SB24-75. Released a policy memo and blogpost.
Learn More | See Blog -
Co-author of a report to Colorado Independent Drivers United (CIDU)
Using rideshare trip data collected via the FairFare tool to measure platform take rate
Read Report -
Co-author of CITP's Supreme Court Amicus Brief in Gonzalez v. Google
Examining scope of Section 230 of CDA for recommender systems.
View Brief
Service
Reviewer / Program Committee Member
- NeurIPS 2024 Ethics Reviewer
- ACL Rolling Review (ARR) 2024
- FAccT 2024
- ACM Creativity and Cognition 2024
- NAACL 22-24
- EMNLP 22-24
- ACL 22-24
- SIGCSE 2023
- WebConf 2023
Invited Talks
-
"Making Rideshare Governable"
National Employment Law Project (NELP), Sept 2024 | [Slides]
-
"On Ad Delivery Algorithms"
Course: Ethics of Computing (Prof. Arvind Narayanan), Fall 2023, Princeton University
-
"Societal Impact of AI on Labor - Studies of Rideshare and Social Media Ad Platforms"
Course: Designing the Future of Work (Prof. Andrés Monroy-Hernández), Spring 2024, Princeton University
-
"Societal Concerns in Targeted Ad Delivery Algorithms"
Course: Responsible AI in Societal Deployment (Prof. Lydia Liu), Spring 2024, Princeton University
Workshops and Summer Schools
-
Workshop on Worker Data Collectives to Improve Accountability, Combat Surveillance and Reduce Inequalities [Organizer]
ACM CSCW, Costa Rica, Nov 2024
Workshop Website | Paper -
Who Counts? Sex and Gender Bias in Data - Workshop [Participant]
University of California Los Angeles, July 2022
-
Graduate Summer School on Algorithmic Fairness [Participant]
University of California Los Angeles, July 2022
-
Boston Differential Privacy Summer School [Participant]
Boston University, June 2022
Mentorship
I am fortunate to work or have worked closely with many excellent undergrad students at Princeton:
- Cathy Di
- Abani Ahmed
- Kyler Zhou
- Eesha Agarwal
- Kristoffer Selberg
- Jessica Ereyi
- Kok-Wei Pua