Keynote Talks

Keynote Talks

  • 01

    Anirban Dasgupta

    Professor, Computer Science – IIT Gandhinagar
    Keynote Talks

    Topic: Coresets for efficient machine learning

    Abstract
    In the face of the data onslaught, smart algorithms have a significant role to play. Over the last couple of decades, coresets, a small and efficiently calculable data summary, have grown in popularity, both in theoretical and practical settings. They enable approximating large optimizations while needing only a fraction of the resources. In this talk, we will survey the recent frontier results on coresets, including results on tensor factorization, fair machine learning, and coresets and privacy.

    Bio:
    Anirban Dasgupta is currently a Professor of Computer Science & Engineering at IIT Gandhinagar. Prior to being at IIT Gandhinagar, he was a Senior Scientist at Yahoo Labs Sunnyvale. Anirban works on algorithmic problems for massive data sets, large-scale machine learning, analysis of large social networks, and randomized algorithms in general. He did his undergraduate studies at IIT Kharagpur and doctoral studies at Cornell University. He has received the Google Faculty Research Award (2015), the Cisco University Award (2016), the ICDT Best Newcomer Award (2016), the Google India AI/ML Award (2020), and the ACM STOC 10 year Test of Time award in 2024.

  • 02

    Christina Leslie

    Member & Professor - Computational and Systems Biology Program, Memorial Sloan Kettering Cancer Center
    Keynote Talks

    AI for single-cell and spatial biology

    Abstract
    We will present new AI models for single-cell multiome (scRNA+ATAC-seq) and imaging-based spatial transcriptomics with applications to decoding immune cell state dynamics and the organization of the tumor microenvironment. First, we will present a generative neural ordinary differential equation (ODE) model for single-cell multiome data called DynaVelo, which we use to learn a functional form of the dynamics — corresponding to learned RNA velocity and transcription factor (TF) motif velocity vector fields — of wildtype and mutant germinal center B cells.

    We show that Jacobian analysis or in silico perturbations can recover dynamic regulatory networks governing cell state transitions in the germinal center. We also show how to predict the impact of genetic loss-of-function mutations on cell dynamics, and how to predict TF perturbations that rescue loss-of-function phenotypes. Second, we will present a new discrete representation learning model called Nicheverse for annotating neighborhoods and cells in imaging-based spatial transcriptomics (imST) by exploiting pixel-level transcript information, without using cell segmentation or converting imST data to a cell-by-gene matrix. We apply our model to imST data profiling pancreatic ductal adenocarcinoma (PDAC) tumors from Memorial Sloan Kettering patients to decode the local microenvironment of perineurally invaded versus non-invaded nerves.

  • 03

    Ganesh Ramakrishnan

    Professor, Computer Science – IIT Bombay
    Keynote Talks

    Topic: The Scaling Loop: The Secret Relationship Between Small and Big Models from the experience at BharatGen

    Abstract
    While frugally scalable AI is often attributed to "necessity as the mother of invention" we pick up that thread and ask the harder question underneath it: why does constraint produce better models? BharatGen has built foundational models across 22 scheduled Indian languages at four different scales, under real resource limits, and learned something at each one that changed the next. This talk is about what that sequence taught us, and why the choice between building small and building large turns out not to be a choice at all.

    Bio:
    Prof. Ganesh Ramakrishnan is Institute Chair Professor at the Department of Computer Science and Engineering, IIT Bombay. Ganesh is also the Founding Director of BharatGen, India's first government-backed, open-source, multilingual foundation model initiative launched in October 2024, spanning text, speech, and document vision models across 22+ Indic languages. His research in natural language processing, generative AI, and human-assisted machine learning defines BharatGen's approach: building large language models on small, carefully curated datasets to make high-quality AI viable in low-resource settings.

    After earning his PhD from IIT Bombay in 2004, he joined IBM Research India, working on scalable information extraction, before returning as faculty in 2009. He led research in statistical relational learning and machine translation, producing Udaan, an end-to-end machine translation ecosystem for Indian languages, and DECILE, a data-efficient machine learning framework for resource-constrained environments. He has collaborated extensively with IBM Research, Adobe, and Google, and his work has appeared in AAAI, ACL, NeurIPS, ICML, and EMNLP; he serves as area chair for AAAI and ACL.

    Beyond research, he is Institute Chair Professor in the Department of Computer Science and Engineering, holds the Bank of Baroda Chair Professorship in Digital Entrepreneurship, and heads the Koita Centre for Digital Health at IIT Bombay. He has been globally recognised as one of the five Goalkeeper Champions by the Gates Foundation (2026), and featured among AIM 100 Most Influential Global AI Leaders (2026 and 2024), MIT Sloan India's AI50 (2026), and Forbes India's Top 30 AI Minds (2025). He is also a recipient of the Dr. APJ Abdul Kalam HPC & AI Award (2026) and the National Gold Award for eGovernance (2022).

  • 04

    Manohar Swaminathan

    Senior Principal Researcher, Microsoft Research India
    Keynote Talks

    From Papers to Impact: Reimagining Research in the Age of AI

    Abstract
    The research enterprise has been degrading for many years, well before the advent of generative AI, drifting from the quest to expand humanity's knowledge to the mundane goal of getting papers accepted in a hierarchy of ranked venues. Researchers across all fields are complicit in sustaining this charade, because all our rewards and recognitions are firmly tied to publication metrics. Individually, we have been powerless to change the system.

    AI has done the research enterprise a huge favor by upending the publication process. Even a basic chat subscription now lets anyone produce novel, high-quality papers that would easily have cleared the gates of top conferences a few years ago. The deluge of papers being submitted to top conferences and the clear evidence of the role of AI in causing this flood is a reality check. Given this scenario, research communities such as CODS have a responsibility to rethink the research process. In this talk, I outline how this might be attempted.

    First, the focus shifts to the importance and impact of the work, with researchers defining both and the metrics by which impact is measured. Second, the research journey is documented through the artifacts it generates. On reaching their milestones, researchers submit a curated collection of these artifacts for review, with a short statement on why the work matters. Distinguished researchers, who themselves have done impactful work in the area, evaluate it and, if they find it significant, endorse it and make the collection available to the community. Third, I describe the key role agentic AI tools must play in this process, which sidesteps entirely the debates on authorship and integrity while enabling follow-up work on promising directions.

    Finally, I argue that this process both democratizes and decolonizes research. It also redirects research efforts, predominantly funded by taxpayers, back toward positive impact on human well-being.

    Bio:
    Manohar Swaminathan is a Senior Principal Researcher at Microsoft Research India and co-convener of the Center for Accessibility in the Global South at IIIT Bangalore. His persistent aspiration has been to engineer a better quality of life for marginalized communities. He draws on expertise across education, mobile computing, IoT, HCI, AR/VR and, most recently, AI.

    His career has spanned academia, industry, and the non-profit sector. He holds a BE (Honours) in ECE from Government College of Technology, Coimbatore, an ME in EE from IISc Bangalore, and a PhD in CS from Brown University. He has been a Professor at the Indian Institute of Science, tenure-track faculty at the University of North Carolina, visiting faculty at the University of Missouri, and a visiting researcher at Columbia University. He co-founded the technology enterprises PicoPetaSimputers, Strand Life Sciences, LimberLink Technologies and Escape Velocity Accelerator, as well as the non-profit Metastring Foundation, and has been an angel investor. At LimberLink he ran Jed-i (Joy of Engineering, Design, and Innovation), a program that worked to build a cadre of top-notch engineers in India at scale.

    For the past three years at MSR India, he has led Project Ekalaiva (Education, Knowledge-skills Acquisition and Livelihoods in the age of AI). The project uses AI itself as a powerful ally in the radical transformation needed in India’s education and research ecosystem. He writes a series of articles on this theme on Medium. He is also building Agentic Shiksha, the platform needed to support this transformation at scale. Its agentic Teaching Assistant, a core component, was recently open sourced on GitHub.

  • 05

    Nisheeth Vishnoi

    Professor, Computer Science - Yale University
    Keynote Talks

    Topic: AI, Work, and Human Ability

    Abstract
    How does AI change the role of human ability in work, and what happens to that ability when AI is used repeatedly? In this talk, I will use two mathematical models to study these questions. The first examines how human abilities in execution and verification shape decisions to delegate work to AI, and how AI changes the relative economic value of these abilities in the short term. The second studies the co-evolution of delegation behavior and human ability in the long run.

    Bio:
    Nisheeth K. Vishnoi is the A. Bartlett Giamatti Professor of Computer Science at Yale University, where he co-founded the Computation and Society Initiative. His research spans theoretical computer science, optimization, and artificial intelligence, with current interests in AI alignment and safety, human–AI interaction, and the societal and institutional consequences of AI. Previously, he held positions at EPFL, Microsoft Research India, and IBM Research India. He is a Fellow of the ACM, AMS, and IEEE, and has received several research awards, including the 2025 FOCS Test of Time Award and best paper awards at FOCS and ACM FAccT. He received his undergraduate degree from IIT Bombay and is the author of Algorithms for Convex Optimization (Cambridge University Press, 2021).