I am a researcher in public interest technology, human-centred data science, and AI governance at the University of Toronto. I work with child welfare agencies, hospitals, and other public institutions to design, audit, and govern AI systems in the public interest.

About:

I am an Assistant Professor in the Faculty of Information at the University of Toronto, cross-appointed to the Department of Computer Science, and I direct the Human-Centred Data Science Lab. My work sits at the intersection of public interest technology, human-centred data science, and AI governance. I study how algorithms and data infrastructures are designed, deployed, and audited inside public institutions such as child welfare agencies, hospitals, and other parts of the state.

I am fundamentally trained as a statistician, with extensive cross-training in science and technology studies, sociology, and economics. That background shapes how I work: I combine technical approaches in AI and machine learning with qualitative and interpretive methods to understand how risk, prediction, and measurement become entangled with power, discretion, and institutional practice. A central focus of my research program is to build and evaluate tools, frameworks, and methodological standards for auditing and governing AI systems so that they are accountable to the communities they affect, rather than only to technical or commercial goals.

Much of my research is grounded in long-term collaborations with public-sector partners in child welfare, healthcare, and related domains. Together, we co-design data systems, predictive models, and decision support tools, and we study how they actually work once they are embedded in everyday organizational life. The aim is not simply to critique harmful systems from a distance, but to reshape how public institutions understand and use AI, and to create infrastructures, audit practices, and governance mechanisms that make algorithmic decision-making more transparent, contestable, and just.

I am also deeply invested in teaching and pedagogy. I coordinate the Human-Centred Data Science concentration in the Master of Information program and co-authored Human-Centered Data Science: An Introduction (MIT Press, 2022), which frames data science as a fundamentally sociotechnical practice rather than a purely computational one. Through my courses, the HCDS Lab, and open educational resources and toolkits, I am building curriculum and training pathways that prepare students and practitioners to design, evaluate, and govern data-driven systems in the public interest.

Across these strands, my long-term goal is to help define a rigorous, institutionally grounded agenda for public interest technology: a research and practice ecosystem in which AI and data systems are engineered, studied, and governed with explicit attention to social justice, public accountability, and the realities of bureaucratic life.

Academic Training and Research Interests:

I am fundamentally trained as a statistician and machine learning researcher, with significant cross-training in science and technology studies, sociology, and economics. I completed my PhD in Information Science and Statistics at Cornell University, then spent five years as an Assistant Professor of Computer Science at Marquette University before joining the University of Toronto in 2021. That trajectory means I think about data and algorithms simultaneously as mathematical objects, organizational tools, and sites of power.

My research program sits at the intersection of computing and the social sciences. Methodologically, I integrate technical approaches in statistics, AI, and machine learning (for example, generalized linear models, predictive modelling, and causal inference) with qualitative and interpretive inquiry (ethnography, interviews, document and content analysis). I also draw on methods and concepts from human–computer interaction and computer-supported cooperative work to understand how data systems are actually taken up in practice.

Substantively, my work is organized around several linked themes:

Public interest technology and AI governance in public institutions
How child welfare agencies, hospitals, and other public organizations adopt, resist, and reshape algorithmic systems, and how we can design audit and governance practices that make these systems more accountable.
Human-centred data science methods and infrastructures
Conceptual and methodological frameworks for “human-centred data science,” including mixed-method audit protocols, evaluation standards, and tooling that can be adopted by practitioners and regulators.
Risk, measurement, and bureaucratic decision-making
How notions of risk, prediction, and performance metrics travel into frontline work, and how they interact with professional judgment, discretion, and structural inequality.
Pedagogy and capacity-building in public interest technology
Curriculum, textbooks, and open educational resources that train students and practitioners to design, evaluate, and govern data-driven systems in the public interest.

Together, these strands form a coherent research program aimed at building a rigorous, institutionally grounded foundation for public interest technology and human-centred data science.

I have been extremely fortunate to have my work recognized through a few distinctions throughout the years such as a Way-Klingler Early Career Award’19, Connaught New Researcher Award’22 and Schwartz-Reisman Institute for Technology and Society Faculty Fellowship’23 as well as several Best Paper/Honorable Mention/Impact Awards at premier CS conferences.

Policy, Practice and Public Interest Impacts

A central goal of my work is to ensure that research on public interest technology and human-centred data science has concrete consequences for how public institutions actually operate. Much of my research is therefore embedded inside real organizations rather than conducted at arm’s length.

I maintain long-term collaborations with public-sector partners in child welfare and healthcare, where I work alongside practitioners, administrators, and technical staff to design, evaluate, and audit data-driven systems. This includes developing decision-support tools, risk assessment workflows, and AI audit practices in ways that fit the messy realities of frontline work, rather than assuming idealized conditions or purely technical objectives.

Beyond individual projects, I strive to produce artefacts that can travel: toolkits and templates for AI auditing, documentation practices for data systems, and training materials for public servants, students, and practitioners. These materials are designed to be adopted and adapted by child welfare agencies, hospital systems, and other public organizations that are trying to govern AI in a responsible, accountable way.

I also contribute to policy and standards conversations around data and AI in the public sector. This has included co-authoring policy reports, participating in advisory work with government and non-profit organizations, and helping to shape curricula and guidance for international initiatives focused on responsible and trustworthy AI. Across these activities, my aim is consistent: to build bridges between critical scholarship, technical practice, and institutional decision-making so that AI systems in public life are more transparent, contestable, and oriented toward the public interest.

Speaking, Consulting and Media:

I regularly give talks and workshops for academic, public-sector, and industry audiences on public interest technology, human-centred data science, and AI governance. My talks often focus on how algorithms and data systems reshape decision-making in child welfare, healthcare, and other public institutions, and on what it takes to design and audit these systems in ways that are accountable to the people they affect.

In my consulting and advisory work, I collaborate with public agencies, non-profit organizations, and policy teams that are deploying or evaluating AI and data-driven tools. This includes helping organizations assess proposed systems, design governance and audit processes, interpret empirical findings, and build internal capacity to work with data and AI responsibly.

I also engage with journalists, professional associations, and civil society organizations that are interested in the societal implications of AI in public life. If you are interested in inviting me for a talk, workshop, panel, expert consultation, or media interview, please contact me by email with a brief description of your organization, audience, and goals.

The best way to reach me is via email { shion [dot] guha [at] utoronto [dot] ca }; I am usually quite responsive to such inquiries, especially those that are time-sensitive.

For more information, please see my google scholar profile for an updated list of publications. I do not use any social media.

For Prospective Students and Trainees:

I work with students and trainees who are serious about public interest technology, human-centred data science, and AI governance in public institutions. If you are primarily interested in generic “AI for good” branding, adtech, or optimizing engagement for platforms, we are probably not a good fit. If you care about how data and algorithms shape child welfare, healthcare, and other public services, and are willing to do slow, careful, sometimes uncomfortable work with real institutions, then we might be.

My group typically includes students from information, computer science, public health, and related disciplines. I am especially interested in trainees who:

Are comfortable crossing technical and social-science boundaries (e.g., statistics/ML plus qualitative or interpretive methods).
Care about institutions and bureaucracy, not just individual users.
Are willing to read deeply, write clearly, and do fieldwork or empirical work that does not always look glamorous on paper.
Want to build things i.e methods, toolkits, frameworks, curricula, that can be used by practitioners, policymakers, and communities.

If you are a prospective Master’s or PhD student at the University of Toronto and are interested in working with me, please:

Read the “About” and “Academic Training and Research Program” sections of this page.
Look at a few of my recent papers to see whether the questions, methods, and writing style resonate with you.
Email me with a brief note explaining:
- How your background connects to public interest technology / human-centred data science.
- What kinds of problems you want to work on (as concretely as you can).
- A CV and, if possible, a short writing sample.

I unfortunately cannot respond in detail to every inquiry, but I do read these messages and I am much more likely to reply when it is clear that you have taken the time to understand what my group actually does and why you want to be part of it.

Past Lives (Before UofT):

Before moving to my current position at the University of Toronto in 2021, I was an Assistant Professor in the Department of Computer Science at Marquette University (2016–2021). Prior to that, I completed a PhD in Information Science and Statistics at Cornell University, where I worked with Steve Wicker on the technical and social dimensions of location-based social networks. Even earlier, I studied at the Indian Statistical Institute, working with B. S. Daya Sagar on mathematical morphology and spatial networks. Those experiences cemented my interest in bringing together rigorous quantitative methods with critical thinking about institutions, infrastructure, and power.

Long before I imagined an academic career, I was convinced I would be a musician. I spent years playing keyboards and keytar, first in the Athens music scene and later in the resurgence of Bangla rock. I did not, in the end, become a rock god, but that part of my life still shapes how I think about collaboration, performance, and craft and I remain happiest when there is a good band on stage and too many cables on the floor.