I work broadly within the realm of gravitational-wave astronomy. Nearly infinitesimal ripples in the fabric of spacetime, gravitational waves are generated by the most cataclysmic events in the Universe, including the explosions of stars and the relativistic collisions of black holes. My interests lie in the use of gravitational waves as tools with which to explore the binary neutron star and black hole population, probe the astrophysical and cosmological gravitational-wave backgrounds, and test our understanding of fundamental physics. Practical gravitational-wave astronomy began only in 2015 with the first successful detection of a binary black hole collision by the Laser Interferometer Gravitational-Wave Observatory. Since then, however, the field has grown exponentially; roughly one hundred gravitational-wave signals have been measured to date, and thousands more are anticipated in the coming years. This expected torrent of data promises to unlock answers to a vast number of astrophysical and cosmological questions, but it will also stress current data analysis methods to their breaking points. My goal as a Schmidt fellow is to identify, develop, and deploy machine-learning methods that will tame this flood, increasing the efficiency and accuracy with which we can analyze gravitational-wave signals and study the properties of their astrophysical progenitors.

BIO:

I grew up in Walla Walla, WA, after which I received a B.A. in Physics & Astronomy from Carleton College, an M. Phil. in Astronomy from the University of Cambridge, and my Ph.D. in Physics from Caltech. I spent three years as a research fellow at the Flatiron Institute’s Center for Computational Astrophysics, after which I moved here to Chicago!

Website

Cellular dysfunction in human disease occurs in a complex, three-dimensional space. However, many state-of-the-art techniques for studying cells rely on the dissociation of tissue to parse cells one by one, destroying any spatial context about the cellular environment. Using novel staining and imaging techniques, we can now probe the cellular constituents of human tissue with higher phenotypic resolution than previously possible and capture over 40 cell markers in a single biopsy, while preserving the spatial organization of cells in tissue. Artificial intelligence (AI) is necessary for high throughput processing of this data, as humans have difficulty interpreting information from multiple imaging channels. In my work, I develop and implement various AI methods for quantitative analysis of highly multiplexed microscopy images. Using AI, we can automatically find and characterize cells in high-content image data and extract spatial features describing the cellular organization of disease states. These spatial features might also be linked with clinical features such as therapy response or patient prognosis. I am particularly interested in exploring the spatial context of immunity in various pathogenic states, ranging from autoimmunity to cancer.

BIO:

Madeleine is a postdoctoral researcher in the Department of Radiology at the University of Chicago. She received her bachelor’s degree in Biomedical Engineering from Vanderbilt University in 2013 and her PhD in Biomedical Engineering from Texas A&M University in 2018. Her doctoral thesis focused on radiative transport modeling to help inform the design of optical imaging and sensing systems to detect disease in vivo. As a postdoc, she works with AI to quantify high-content optical microscopy images of human tissue samples. She is also interested in using AI to merge data from multiple imaging modalities to improve predictive models. In her free time, she likes to get away from the computer and run, hike, or walk; anything to stay active!

Meeting the surging energy demands calls for the holistic development of batteries with significantly higher energy densities than the contemporary Li-ion batteries. Lithium metal batteries (LMBs) offer up to ten times higher energy densities and therefore represent a promising alternative. However, most electrolytes developed for LMBs have poor compatibility with the lithium metal, which is highly reactive. Despite extensive Edisonian experimental investigations over the past five decades, none of the discovered electrolytes can sustain desirable long cycle lives. My research efforts involve developing and utilizing forward design (structure to property mapping) interpretable ML frameworks, Bayesian optimization (adaptive design) of targeted properties of electrolytes, and inverse design (property to structure mapping) of electrolytes with desired properties using generative algorithms. These AI/ML approaches will enable exploration of the electrolyte chemical space on an astronomical scale that remain vastly unexplored through traditional scientific methods.

BIO:

I joined UChicago as a postdoctoral fellow at the Pritzker School of Molecular Engineering under Prof. Chibueze Amanchukwu in 2022. I received my doctoral degree (Ph.D.) in materials science from Indian Institute of Science, India in 2022, with my dissertation formulating robust strategies for designing efficient materials using quantum chemistry-based methods and ML algorithms. My research interests span wide areas of materials informatics aimed at developing novel materials, at the intersection of physics, chemistry, materials science, computational modeling, data science, and cheminformatics. I also hold a M.S. in Chemical Sciences from the Indian Institute of Science (2017) and a B.Sc. in Chemistry (Honors) from the University of Delhi, India (2015).

Website

RESEARCH:

Our goal is to design sustainable polymeric materials through combining material science, design, and engineering with AI. This involves three main objectives, including understanding the impact of monomeric chemistry, representing polymers as variable ensembles, and developing faster integration schemes for simulations, ultimately aiming to design new materials for the circular economy.

My scientific background is focused on material science, design, and engineering, using MD simulations, statistical mechanics, and chemical informatics. Through the Schmidt Foundation for AI + Science, I am combining my expertise with AI to design sustainable polymeric materials and gain a deep understanding of the dynamics, rheology, and morphology of high-performing battery and membrane materials.

The scientific goal of our research is to develop a fully in silico workflow that enables the design of sustainable polymeric materials, addressing the urgent problem of plastic pollution and contributing to the circular economy. In silico screening and design are essential to make this transition economically feasible.

My mission encompasses three main objectives:

1. To achieve fundamental sustainable material design, we need to understand the impact of monomeric chemistry. To achieve this, we propose high-throughput atomistic simulations combined with active learning to explore and understand the vast chemical space. A polymer-centric database is essential, and I contributed to building CRIPT, a scientific FAIR data bank for polymer materials.
2. Representing polymers as variable ensembles of macromolecules is necessary for any machine learning method to work. Encoding the ensemble and chemical specificity is a formidable challenge, and we propose a new Ansatz that combines local chemical information with the molecular graph ensemble.
3. Developing faster integration schemes for simulations is essential. We propose to develop a kinetic Monte Carlo integrator that fulfills detailed balance irrespective of its trained state.

Our ultimate goal is to design new materials that are sustainable and contribute to the circular economy. Through our interdisciplinary approach, we are taking significant steps towards achieving this goal.

BIO:

Ludwig is a computational material designer specializing in polymeric and soft matter. With a background in physics, he has combined his expertise in software engineering, chemistry, and data science for numerous research projects. Ludwig completed his undergraduate and graduate studies under the guidance of Prof. Müller at Georg-August University in Goettingen, Germany, where he studied rheology and structure formation in complex polymer melts.

After receiving his Ph.D. degree, Ludwig moved to Juan de Pablo’s group at the Pritzker School of Molecular Engineering, University of Chicago, to broaden his skill set to include machine learning and chemistry. While at the University of Chicago, Ludwig’s research focus shifted towards Machine Learning and AI, where he developed automated simulations to explore the chemical space and find more sustainable plastic materials. He also developed methods to accelerate the time evolution prediction of high functioning nano-materials.

As an AI in Science Fellow of the Eric and Wendy Schmidt Foundation, Ludwig’s primary goal is to build AI tools and promote the use of AI in Science. A prime example of his work is the creation of CRIPT, a data bank for polymeric materials that serves as a foundation for machine learning in polymer science.

Website

RESEARCH 

As a Schmidt AI in Science Fellow, I will leverage novel AI/ML approaches to understand one of the most basic questions in biology – how cells and tissues function collectively and how this knowledge can be exploited to engineer living systems. Specifically, focusing on the role of endothelial mechanotransduction, I will utilize ML algorithms to predict the cell and tissue-scale dynamics contributing to health and disease, and engineer model tissues that mitigate disease progression.

BIO:

Shailaja Seetharaman is a postdoctoral researcher in the Gardel Lab at the University of Chicago. She received her PhD in Cell and Developmental Biology from Institut Pasteur, Paris, and her Master’s in Biomedical and Molecular Sciences from King’s College London. Her research interests lie in the field of cell and tissue mechanics, focusing on the role of cytoskeletal networks and adhesion complexes in physiology and pathology. In particular, she is working towards understanding endothelial mechanotransduction in disease progression. She is the recipient of American Heart Association Postdoctoral Fellowship, Yen Postdoctoral Fellowship, and the Marie Curie and FRM PhD fellowships.

RESEARCH:

To understand the neural computations that lead to object understanding and guide flexible, naturalistic behaviors, I deploy AI techniques to inform efficient visual neuroscientific experiments. Using closed-loop neuroscientific experiments (AI generates hypotheses -> experimental data informs AI models) I will test the central hypothesis that learned associations between object-scene properties affect the inference of those properties to guide behaviour.

BIO:

My research focuses on understanding how the brain processes visual information to guide flexible behavior. I use electrophysiological, psychophysical, and computational techniques to study how the primate visual system processes objects (presented on a screen) and how inferences about those objects are mapped to behavioural outputs in different environmental, cognitive, and task conditions. During my Ph.D. in the labs of Drs. Ed Connor and Kristina Nielsen, I studied how 3D object information is extracted from 2D images using single-unit extracellular electrophysiology and two-photon imaging in monkeys. I brought my experience with algorithms to generate parameterised, naturalistic 3D visual stimuli to my postdoc in the lab of Dr. Marlene Cohen. The broad goal of my research programme is to understand how our visual brain enables us to interact flexibly with our world by inferring relevant properties of 3D objects in naturalistic environments.

Website

My research seeks to use simulation tools to better understand biological systems. There are several challenges when using molecular simulations to study biological systems. One is how to represent a real system accurately with a model. Another challenge is that many events of interest are rare. That is, even with the most powerful computational resources currently available, it can still take hundreds of years or more to simulate molecular processes. Finally, it is also a challenge to understand the huge datasets generated by molecular simulations. I try to address these issues by combining artificial intelligence (AI) with statistical mechanics. Statistical mechanics provides the theoretical basis for the simulation study of biological systems, while AI has shown great potential in automatically making sense of large, high-dimensional data. Their combination opens up possibilities for overcoming the curse of dimensionality and bridging the gap between theory and applications/simulations. Therefore, I hope to utilize their synergies to make simulation tools applicable to systems with arbitrary complexity and timescale.

BIO:

Yihang received his B.S. degree in Physics from the Southern University of Science and Technology (SUSTech) in 2017.  He then started his Ph.D. studies in the Biophysics program at the University of Maryland, College Park (UMD). Advised by Prof. Pratyush Tiwary, he explored the possibility of combing AI techniques with molecular dynamics simulation to study the interaction mechanisms of (bio)molecules.He was a postdoctoral fellow at the Chicago Center for Theoretical Chemistry (CCTCh) before becoming an Eric and Wendy Schmidt AI in Science Postdoctoral Fellow.

Website