I study how new technologies change work and organizations, especially when they move from being exciting tools to becoming part of how work actually gets done.
I'm a PhD candidate in the Management of Organizations (Macro) group at UC Berkeley Haas School of Business.
Right now, I am focused on generative AI. I am interested in how AI changes productivity, workflows, evaluation, expertise, and organizational design. I care about the practical questions organizations are facing: when AI helps people do better work, when it creates new kinds of work, how firms should structure knowledge so AI systems can use it, and how managers should govern AI-assisted or AI-executed work.
At the same time, I do not think technology adoption is ever just a technical problem. The same tool can have very different effects depending on how it is introduced, who uses it, what work it is applied to, and how organizations evaluate its outputs. That is what makes this moment so interesting: generative AI is not just changing individual productivity; it is pushing organizations to rethink roles, workflows, accountability, and the infrastructure of knowledge work.
Before starting the PhD, I spent six years working as a data scientist.
Some of the questions that motivate my research are:
- How is AI changing collaboration and coordination at work?
- How do organizations build the infrastructure needed to use AI at scale?
- What should be transparent when AI is involved in work?
- Who benefits from new technologies, and who gets left out?
Across these questions, my research examines how organizations adopt, adapt to, and are changed by new technologies.
-
Not Worth the Effort: Technology and Effort Opacity in Collaborative Work
Job market paper
Generative AI (AI) makes individuals more productive, but work is often collaborative: one person creates work, and another reviews it. Collaboration therefore relies on a mutual exchange of effort. Before AI, effort was often inferred from the work itself. I argue that by making work cheaper to produce, AI creates effort opacity—it makes it harder to infer the human effort behind the work. This change in the informational property of work can reduce how much review AI-assisted work receives. I document this change in collaboration using observational data from top projects on GitHub that have over 1 million stars and 20 million dependencies. I find that even after controlling for quality, AI code receives 25% less detailed but 22% more general feedback, is more likely to be rejected and therefore not integrated into the project. I then test whether perceived effort drives this reduction in review in a preregistered experiment with software developers. I find that when effort cues are weak, AI-disclosed work receives less review effort compared to human work. However, when effort cues are strong, AI-disclosed and human work receive similar review effort. These findings show that AI can make work easier to produce while making it harder to attract the feedback needed to improve and integrate it. As AI adoption increases, organizations may need new ways to make human effort legible in work.
-
Which Ideas Attract Workers? Gender and Labor Supply in Startups
Revise & resubmit Management Science
Which startups can attract the talent they need to grow? We argue that worker preferences over which problems to solve shape labor supply to early-stage firms. In male-dominated labor markets, startups focused on women may struggle to recruit, especially for critical, male-dominated roles like engineering and senior leadership. We combine observational data on over 52,000 U.S. high-potential startups with a pre-registered field experiment. Linking startups to over 2.5 million worker histories, we find that female-focused startups employ fewer workers: a one standard deviation increase in a startup's female-focus predicts 11% fewer male employees. To isolate worker preferences, we run a field experiment where we randomize the startup idea itself, thereby developing a new experimental paradigm to causally identify how the nature of an idea matters. We built a job-search platform recruiting real tech job seekers in a within-subject design, where each participant evaluates 15 startup ideas, each randomized to be more or less female-focused. Crucially, our experiment is incentive-compatible, because participants' evaluations affect which real job postings our platform subsequently recommends to them. Male job seekers are 6.8 percentage points less likely to apply to female-focused startups than other startups, a 10% decrease from baseline. Offsetting the male penalty would require approximately $47,000 in additional yearly compensation. Our findings demonstrate that who benefits from an idea is an important nonpecuniary job attribute shaping talent acquisition, potentially helping explain why ventures serving women grow more slowly.
Nominated for the Best Conference Paper Award and the Research Methods Paper Prize, Strategic Management Society Annual Conference 2023.
-
Whose Transparency Is It Anyway?
Working paper
Transparency has become a central principle in generative AI governance, yet there is little agreement about what transparency should actually mean, for whom, and in what contexts. Drawing on the Social Construction of Technology and Value Sensitive Design, this paper examines how different stakeholders understand and experience transparency in generative AI. We use a two-phase mixed-methods study: a survey of 400 end users, AI builders, and policymakers, followed by participatory design sessions with end users. We find that stakeholders broadly agree that transparency matters, but differ in what kinds of transparency they need and whether those needs are currently being met. In particular, existing transparency tools appear to serve builders more effectively than end users, revealing how power shapes which interpretations of transparency become embedded in practice. We argue that meaningful AI transparency requires moving beyond one-size-fits-all disclosures toward participatory, pluralistic transparency infrastructures that reflect the needs of diverse stakeholders.
-
Redesigning Organizations Around AI: Roles, Knowledge Infrastructure, and Governance
Work in progress
Generative AI is usually introduced as a tool for individual productivity, but using it at scale is a problem of organizational design. When AI systems begin carrying out tasks that people used to do, firms have to revisit the basic design questions: how work is divided, who does what, where decision rights sit, and what has to be documented for the organization to function. This project examines how a technology company undergoing an AI-first transformation redesigns itself around AI. We focus on three design choices. First, the division of labor: how roles change when employees become the people who build, direct, and supervise AI-enabled work rather than perform it, and what happens to the entry-level tasks through which people used to learn the job. Second, knowledge infrastructure: what an organization must make explicit and legible before AI systems can act on it, and how the work of documenting and curating that knowledge gets allocated. Third, governance: what oversight structures, accountability, and decision rights are needed when AI systems execute organizational tasks. By treating AI adoption as organizational redesign rather than tool use, this study contributes to research on organizational design by showing how firms rebuild roles, knowledge, and authority around a general-purpose technology.
I learn the sitar under the guidance of my guruji, Rajib Karmakar. Here's a short clip of one of the events I performed at.
An old habit of reading widely and writing to think. For a while I ran a newsletter rounding up recent management research, and wrote data-driven essays on cities, culture, and everyday life. I haven't posted in a while, but I like keeping the archive around.
I'd love to chat: about AI and work, research, or anything adjacent. The best way to reach me is email.