My primary research investigates internet access disparities in the U.S. through a socio-economic lens. Despite significant technological progress, substantial gaps in internet quality persist across neighborhoods — gaps that are usually chalked up to infrastructure investment, but rarely examined with much demographic nuance or geographic precision beyond urban vs. rural.

My work fills that gap by stratifying urban areas by population density and using Ookla's crowdsourced speed tests for their broad coverage. I tackle the sampling biases inherent to crowdsourced data and apply regression models over temporal datasets to surface trends that hold. The takeaway: internet inequality is locally configured — there's no single national narrative, and effective policy needs region-specific intervention.

This work is done in collaboration with Prof. Elizabeth Belding at UC Santa Barbara.

Read the full paper here.

While my main research deals with answering ”who gets what quality of internet”, another line of research investigates ”what those speeds actually translate to in practice”. I am collaborating with my teammates to examine how application-level performance for web browsing and video streaming varies with respect to Ookla measurements — identifying when such measurements are reliable predictors of user experience and at what point they stop mattering.

As part of this effort, I am leading the collection and analysis of YouTube and YouTube Shorts performance metrics to evaluate how these indicators respond to varying network conditions. This work seeks to bridge the gap between network-level measurements and real-world application performance, offering a more user-centric understanding of internet quality.

This work is done in collaboration with Humaira Fasih Ahmed Hashmi at UC Davis.

My secondary interests lie in the security and privacy of internet infrastructure. Public services increasingly live on the web, and citizens depend on these sites for essential tasks — often with no practical alternative. The central concerns are:

• who can observe or process data in transit
• where it is routed and stored
• which jurisdiction governs it
• which networks or facilities actually operate the service

Our goal is to produce a rigorously measured, state-wise map of hosting strategies and security posture across U.S. state and federal websites, with explicit accounting of sovereignty and third-party vendor dependence.

This work is done in collaboration with Sachin Kumar Singh at University of Utah.

Measuring the quality of connectivity of libraries, schools, and hospitals at scale is difficult because their IP addresses aren't documented anywhere. This work presents Reverse IP Geolocation (RG), a framework that infers IP addresses from publicly known institutional street addresses, validating candidates against DNS records, WHOIS data, broadband provider data, and active measurements.

My contribution in this project focused on validating that the dataset was geographically representative — examining whether coverage held consistently across urban and rural areas and across individual states for the method to be applicable across the country.

This work is done in collaboration with Nishant Acharya at UC Davis.

Read the pre-print here.