About
I am a postdoc at Princeton's Center for Information Technology Policy (CITP), where I study the societal impacts of algorithmic and AI systems. I audit these systems for privacy, bias, and safety issues to provide actionable insight for design and policy changes. I am interested in developing algorithmic solutions for integrating public-interest safeguards into the design of emerging technologies.
I bring over six years of experience in Responsible AI research and also have a strong background in writing software and developing infrastructure for experimentation. I am experienced in algorithmic auditing, privacy-preserving algorithms, statistical data analysis, and effectively communicating technical insights to legal, policy, and multidisciplinary audiences.
My PhD research focused on developing effective methods for measuring bias in machine learning algorithms used by digital advertising systems on social media. I developed black-box audits that uncovered bias along legally protected attributes in Meta's ad‑delivery algorithms for employment and education ads. These contributions have been featured in the popular press, and contributed to a line of research that led to Meta deploying a debiasing system as part of a legal settlement with the U.S. Department of Justice.
Email: imana at princeton dot edu
Bio
I got my PhD in Computer Science from University of Southern California. I had the wonderful opporunity of being coadvised by Prof. John Heidemann and Prof. Aleksandra Korolova. I was part of USC/ISI's Analysis of Network Traffic(ANT) group and USC CS Theory Group.
Prior to USC, I received my BSc in 2017, also in CS, from Trinity College Connecticut, where I worked on solving computationally difficult problems by leveraging high-performance parallel (using GPUs/CUDA) and distributed (using MPI) computing.
During my college years, I also enjoyed participating in competitive programming and robotics. I have competed in Intelligent Ground Vehicle Competition, Trinity College's Fire-fighting Robot Competition, and ACM-ICPC at the regional level.
