My research interests lie at the intersection of human-computer interaction and generative AI. More specifically, I aim to understand how we should (and should not) design AI tools in a way that complements and augments end-user capabilities, rather than replacing them.
Until recently, I was a Research Fellow at the PROSE team at Microsoft, where I worked on evaluating and designing agentic AI tools for programmers and end users, guided by Dr. Emerson Murphy-Hill, Dr. Gustavo Soares and Dr. Sumit Gulwani. By investigating interactions with agents, I aimed to understand how we can better design them to collaborate effectively with users. I completed my bachelor's degree in Computer Science and Engineering at Indian Institute of Technology Kanpur, where I was fortunate to be advised by Prof. Sruti Ragavan. During this time, I investigated the opportunities and challenges of using LLMs for programming education.
VL/HCC 2026 (To appear)
Designed and tested a prototype for a Plan Mode feature for Excel Copilot (a version of which was released for production!) finding that planning improved user perceptions and reduced effort to refine incomplete solutions.
ASE 2025
An empirical study to understand the strategies developers use when collaborating with AI agents on real-world tasks and the challenges to effective collaborations.
TOCHI 2025
Developed and evaluated a multi-agentic AI tool that assists end-users in creating new spreadsheets.
Pre-print
A mixed methods study comparing how CS2 students learn programming concepts using LLMs versus traditional online methods, examining information gathering approaches and outcomes across both scenarios.
Pre-print
An examination of the cultural biases shown by a variety of LLMs in the symbols that they generate, and experimenting with prompting techniques to reduce such biases.
ACL 2026 (Industry Track)
Defining and evaluating with real users a novel approach to extract tables from semi-structured input text utilizing both neurosymbolic methods as well as LLMs.
SIGCSE 2024
A survey-based study examining how trust in GenAI and the factors affecting this trust influence students' adoption of GenAI and their performance in CS courses.
VL/HCC 2023
An analysis of the participant recruitment biases in end-user programming research and recommendations for more inclusive approaches.
Course Project
Designed a high-fidelity prototype for an app that supports people struggling with binge eating by applying fundamental interaction design principles.
Semester Project, ML4ED Lab, EPFL
Built on the paper by Cock et al. to analyze the impact of intersectionality on different oversampling strategies for a behavioural predictive model in education.
A choose-your-own-adventure style interactive game written in C++ (name inspired from a Netflix Show).