In this monthly interview series, we turn the spotlight on members of the academic community and the important research they do—as thought partners, collaborators, and independent contributors.
For May, we nominated Lav Varshney, an associate professor in the Department of Electrical and Computer Engineering and the Coordinated Science Laboratory at the University of Illinois Urbana-Champaign (UIUC). In this Q&A, Lav shares his experience collaborating with Meta to develop AI-derived formulations for concrete that are less carbon-intensive, and the results he and Professor Nishant Garg, his fellow academic collaborator from UIUC’s civil and environmental engineering department, and the Physical Modeling team at Meta have already seen.
Q: Can you tell us about your background in academia, your role at UIUC, and the type of research you specialize in?
Lav Varshney: I returned to academia in 2014 to work at the University of Illinois Urbana-Champaign, and started collaborating with Meta in 2019. Before this, I was a research staff member at the IBM Thomas J. Watson Research Center, where I led the development of the Chef Watson computational creativity system.
I specialize in the science and engineering of informational systems in modern environments, and draw on several disciplines, including information theory, artificial intelligence, statistical signal processing, data science, sociotechnical systems, and neuroscience. I let my passion for creativity guide my research and enjoy working in diverse fields like food, music, and sustainability.
Q: What have you been working on lately, and how did your collaboration with Meta start?
LV: The work I’m doing with Meta’s Physical Modeling team is one project right now. It’s a project that has a high potential impact on science and on society, and is also fun to work on — three principles I look for in any project I pursue. I started working with Meta in 2019 after my friend Julius Kusuma invited me to a Meta workshop on expanding connectivity. I attended the workshop to contribute to the discussion from a communication-theoretic perspective, but realized the work I’d been doing at UIUC on AI-driven novel building materials could benefit the project.
In addition to my academic work and the project with Meta’s Physical Modeling team, I’m involved in a few AI projects with startups including one on optimizing wastewater treatment. I’ve been interested in climate and sustainability for some time, and I'm working on a new project that aims to use modern AI methods to predict the impact of climate change on agriculture and urban life.
Q: Tell us about the sustainability project you’re working on with Meta’s Physical Modeling team. What are the project goals? What’s your approach for reaching these goals?
LV: I’m collaborating with Meta’s Physical Modeling team to redesign building materials, specifically concrete, so as to have less carbon emissions. Eight percent of worldwide carbon emissions come from the cement in concrete, which is the most widely used engineered material in the world. Reducing these emissions by half, as we may be able to do, will have an incredible impact on the environment.
Our goal is to find new formulations for concrete that are environmentally friendly and have the same strength and performance requirements as traditional concrete. We’ve developed a new AI model that uses variational autoencoders to create concrete that’s sustainable and has the necessary 28-day strength requirements. This work, from computation to field testing, wouldn’t be possible without academia and industry, specifically Meta, working together. Our resources complement each other and have produced results that exceed our expectations.
Q: What results have you seen from the project so far, and where do you see your collaboration with Meta going in the future?
LV: We trained our AI model using a public concrete compressive strength data set and have seen incredible results from the formulas the model created. We’ve been able to deploy low-carbon concrete in the field, which has 40 percent fewer carbon emissions than the regional industry baseline and exceeds construction strength requirements.
We still have a lot of work to do to develop concrete with even less carbon emissions and make it accessible to the general construction industry. We’re focused on data center construction right now because data centers take a lot of energy and resources to build and can be a source of greenhouse gas emissions. It could take a few more years to work through computation and field strength tests that target long-term performance. I’ll be continuing this work in the near term, and in the future, I’d like to explore other research projects in social media, information theory, and sociotechnical systems with Meta.
Q: Where can people learn more about your research?
LV: My university webpage is the best place to learn more about research. You can learn more about the green concrete project on the Tech at Meta website.