“Layerwise Randomized Smoothing for Provably Safe Machine Learning.” 
Brendon Anderson, Assistant Professor, Mechanical Engineering, College of Engineering 

Despite the popularity and impressive performance of artificial intelligence (AI), current AI systems are easily manipulated into yielding catastrophic failures. For example, human-imperceptible attacks on data can cause traffic sign classifiers to predict a stop sign as a yield sign (Kumar et al., 2020), or cause ChatGPT to generate a step-by-step plan to destroy humanity (Zou et al., 2023), or even cause Waymo vehicles to abruptly stop on public roadways (MOTORTREND, 2024). Such susceptibilities prohibit the safe application of AI to safety-critical systems. Randomized smoothing (RS) is a method that intentionally injects noise into data in order to post-process machine learning (ML) models to not only be robust against such adversarial attacks, but also to rigorously certify safety against them. However, prior RS methods result in degraded predictive performance. In this project, we propose a novel, and more flexible, layerwise incorporation of RS into modern ML architectures in order to achieve provable safety guarantees simultaneously with state-of-the-art predictive performance. The project will vastly generalize and improve upon prior works that perform RS in a neural network’s latent space (Pfrommer et al., 2023), and generate new insights into the effectiveness of RS-based methods for AI.

“Embodying Racial Justice: Social Change via (Inter) Personal Transformation.” 
Alexia Arani, Assistant Professor, Women’s, Gender and Queer Studies, College of Liberal Arts 

This project amplifies Black feminist epistemology and anti-racist scholarship by publishing a book of Lace Jeanine Watkin’s racial justice praxis. As an interdisciplinary project focused on public scholarship, this project benefits researchers of social movements, activism, and liberation theory while providing practical guidance for activists, educators, and members of the general public invested in transformative social change.
Lace Jeanine Watkins (1963-2023) was a Black feminist writer, organizer, and educator who drew upon her background in addiction counseling and faith-based organizing to develop a theory of racial justice grounded in relational ethics and psychological transformation. Between 2018-2023, she generated thousands of posts on Facebook (and later, her website) that guided her online followers through the important work of mitigating the harm Black and brown people endure at the hands of white people and white supremacy. This research project preserves and continues Lace’s legacy by curating her educational materials into a book and by producing new knowledge through careful study, synthesis, and analysis of key themes and interventions across Lace’s scholar-activism. In a society defined by “racism without racists” (Bonilla-Silva 2003), this project equips individuals with the tools for interrogating and disrupting the internalization of racial conditioning that supports and maintains systems of white supremacy. In the spirit of community-based research and scholar-activism, the project will benefit Cal Poly and surrounding communities via the training and mentorship of student researchers; integration of anti-racist research into pedagogy, conference presentations, public talks and workshops; and accessible publications.

“Assessing the Efficacy of the BEACoN Research Mentoring Program.” 
Kelly Bennion, Associate Professor, Psychology and Child Development, College of Liberal Arts 

The BEACoN Research Mentoring Program is a university-wide program funded by the Office of University Diversity and Inclusion (OUDI) that supports the participation of underrepresented students in a two-quarter research experience working with a faculty mentor. While the thrust of the program is the one-on-one research experience and mentoring relationship, undergraduate BEACoN Research Scholars attend professional development workshops and community-building events to support their sense of belonging. The time has come to expand the program via external funding; in Fall 2024, the student acceptance rate was less than 10%. However, there is a paucity of data about the program’s effectiveness, which will be necessary for future external grant applications. The current study proposes a multifaceted approach consisting of student focus groups and survey data (with qualitative and quantitative items) from present and past BEACoN Research cohorts (2018-present) to assess the program’s efficacy and its impact on students’ researcher identity, sense of belonging, career trajectories, and more. For example, Researcher Identity, Researcher Confidence and Independence, Equity and Inclusion Awareness and Skills, and Professional and Career Development Skills, as measured by a modified Entering Research Learning Assessment (ERLA) scale, are expected to increase from pre- to post-BEACoN participation for the 2026 student cohort. Overall, this study’s results are critical both in informing the literature on the scholarship of mentoring and undergraduate research – particularly regarding students from marginalized groups – and in collecting data to enhance the program’s effectiveness and therefore better support our students.

“Mobile Wearable Sensor-Based Biomechanical Analysis and Educational Lessons in Elementary and Middle Schools.” 
Britta Berg-Johansen, Assistant Professor, Biomedical Engineering, College of Engineering 

Previous studies have shown that children with overweight, obesity, and decreased physical activity tend to have impaired balance and gait, which can increase risk of musculoskeletal issues in childhood and adulthood. Traditional balance and gait assessment requires a visit to a clinic or a motion analysis lab. Wearable sensors such as inertial measurement units (IMUs) allow for more accessible, inexpensive, and continuous balance and gait measurements that can be used in real-world settings. However, no studies have investigated the effects of BMI and activity levels on balance and gait in children using the IMUs embedded in smartphones and smartwatches. The goal of this project is to use novel wearable sensor-based methodologies to lead interactive biomechanics lessons and investigate the effects of body- and activity-related variables on balance and gait biomechanics in children. This study will promote diversity, equity, and inclusion by using accessible methods and collaborating with schools serving high numbers of underrepresented populations of Hispanic and socioeconomically disadvantaged children. The hypotheses of the study are that (1) data from smartphones and smartwatches may be used to accurately assess balance and gait metrics compared with gold standard methods and (2) balance and gait metrics measured by a mobile biomechanics lab will be significantly associated with BMI, body fat percentage (BF%), and participation in certain activities and sports for children aged 7-14.

“Cellular Multi-Taskers: Characterizing the DNA-based functions of the threonine biosynthetic pathway in Saccharomyces cerevisiae.” 
Jennifer Chik, Assistant Professor, Chemistry and Biochemistry, Bailey College of Science and Mathematics 

As it refers to proteins, the term “moonlighter” was first coined in 1999 to describe proteins with multiple cellular functions. This was considered revolutionary as until then, proteins were assumed to have evolved to accomplish a singular role. When new moonlighters are discovered, they create novel connections between cellular pathways that were previously thought to be functionally distinct. My laboratory focuses on discovering moonlighters that function at the intersection of amino acid metabolism and chromatin biology, two pathways that are absolutely critical to the health and viability of any organism. In the budding yeast Saccharomyces cerevisiae, the amino acid threonine is synthesized from another amino acid, aspartate, by a well-established team of enzymes. During my postdoctoral work, I identified and characterized the first moonlighting roles for enzymes in this pathway in the maintenance of ribosomal DNA silencing. Because DNA-based functions are broad and multi-faceted, I hypothesize that these enzymes have additional moonlighting roles in DNA repair processes. The proposed aims specifically focus on elucidating the DNA repair roles of two pathway enzymes, aspartate kinase (Hom3) and aspartic beta semi-aldehyde dehydrogenase (Hom2). Because the threonine biosynthetic pathway is not found in humans, but is well-conserved in fungi and bacteria, a comprehensive view of Hom3 and Hom2 functional capabilities furthers the argument for their use as attractive microbial-specific pharmaceutical targets. This work will engage students from multiple departments in the Bailey College of Science and Mathematics and will serve as the foundational data used in peer-reviewed publications and applications to external funding.

“Assessing Antimicrobial Biosynthetic Potential from a Library of Ancient Microbes.” 
Rachel Johnson, Research Fellow, and Katharine Watts,  Associate Professor, Chemistry and Biochemistry, Bailey College of Science and Mathematics

Bacteria, fungi, and plants are producers of small molecules called natural products that often have therapeutic potential. In fact, the majority of antibiotic compounds used clinically are natural products or close structural derivatives. The goal of this research project is to identify new antibiotic compounds and to understand how they are made (i.e. biosynthesized) in the natural producer.

Professor Emeritus Raul Cano gifted us a library of ancient bacterial strains isolated from amber and deep-sea core sediments that have yet to be tapped for their chemical and biosynthetic potential. We hypothesize these ancient isolates will have the enzymatic capabilities to produce novel natural products with therapeutic potential. To guide our discovery efforts, we will isolate and assess the full genome sequences of these ancient microbes to search for the genes encoding enzymatic machinery that biosynthesizes antibiotic natural products (Aim 1). We aim to screen the collection of natural products produced by these ancient bacteria for antimicrobial activity against safe relatives of human pathogens (Aim 3). Additionally, we hope to characterize how these molecules are made by their biosynthetic enzymes, which may uncover unique chemistry conducted in nature (Aim 2). Taken together, we can begin to understand what natural products are made by these bacteria and how they are made. Ideally, we will isolate novel antibiotic compounds through extraction and purification methods. By assessing the suite of compounds these ancient microbes produce and how they are synthesized enzymatically, we can better understand how bacterial chemical defenses have evolved over time.

“San Francisco’s Adapting Waterfronts Past, Present, and Future: Visualizing Climate Change Resilience with Augmented Reality.” 
Gabriel Kaprielian, Assistant Professor, Architecture, College of Architecture and Environmental Design  

Imagine a postcard from the future, San Francisco in the year 2100. What might someone write back to the present day and what imagine would illustrate how the city and world have responded to the threat of rising sea levels on the urban waterfront? This project explores the use of augmented reality and mobile web-based tools to visualize climate change resilience. The objectives of the research project are to address the challenges of coastal flooding caused by climate change by sharing knowledge and empowering informed action, enabling discourse between diverse stakeholders and inspiring collective optimism, and fostering transdisciplinary collaboration and international cooperation this global issue. The program will be developed as three interconnecting parts: (1) Research of San Francisco’s Waterfront Past, Present, and Future, (2) Development of Augmented Reality Installations and Public Engagement, (3) Integration into an Interactive Mobile Web-based Tool. The project benefits include engaging public awareness of sea-level rise and potential adaptation solutions through place-based installations that leverage augmented reality technology and the development of an accessible web-based platform for wider dissemination that will enable further collaboration of the research project. The work of the project is significant for San Francisco in advancing awareness of the shared environmental and health and safety risks posed by sea-level rise along with potential solutions. It offers a model for cities around the world to utilize emergent technology in augmented reality that allows for more engagement and productive discourse between built environment professionals, scientists, engineers, city planners, and the public.

“Examining the Convergence between Design Thinking, Critical Thinking, and Technical Professional Communications.” 
Rebekah Maples, Lauren Beck and Erin Martin-Elston, lecturers, English, College of Liberal Arts 

Through vital research and a comprehensive survey of existing university technical professional communications (TPC) programs, we will forge a groundbreaking new scholarly work that examines the unexplored convergence between critical thinking, design thinking, and TPC. Our proposal seeks to create a new pedagogical framework that will advance the field of TPC and pioneers a new approach. While design thinking has become a foundational aspect of TPC, praised for its focus on user-centered design and practical problem-solving, critical thinking—vital for fostering reflection and deep analysis—has been essentially ignored in this context. This project aims to bridge that gap and reimagine what TPC pedagogy can accomplish. Through this work, we also plan to mentor three student researchers who would gain valuable experience in the research and publication process and exemplify Cal Poly's Learn-By-Doing education. By conducting a detailed review of existing scholarship and gathering data on TPC course structures across a range of colleges and  universities, our team will fully examine the convergence between critical thinking, design thinking, and TPC to produce a work of crucial scholarly literature and foster connections between theory and practice.

“Digital California Heritage Lab: Virtual Museum Pilot Project.” 
Elizabeth Minor, Assistant Professor, Social Sciences, College of Liberal Arts  

The Digital California Heritage Lab: Virtual Museum Pilot Project will create an immersive online exhibit that explores the diversity of the state, and preserves and shares artifacts from California museums as 3D assets. This pilot project will provide a proof-of-concept prototype for the National Endowment for the Humanities (NEH) Digital Projects for the Public grant program, with an intended submission date of June 2026. The PI, Dr. Elizabeth Minor, is a new Assistant Professor of Anthropology in the Social Sciences Department, with a focus on Museum Studies and UX (User Experience). Her experience brings together 3D modeling, museum exhibition design, archival research, and visitor/user experience assessment studies. She will recruit undergraduate research assistants from Social Sciences, and related departments, to participate in the 3D capture, virtual gallery design, label researching and writing, and visitor/user experience assessments. The end product will be a publicly accessible virtual museum gallery highlighting a selection of artifacts from local and other California museums. This virtual gallery will be used in Dr. Minor’s courses for in-class Learn By Doing assignments (such as Museum Anthropology and Digital Anthropology), as well as by collaborating museums in their galleries, websites, and social media. The project will provide a key learning resource for California elementary schools, especially for 4th grade state history curricula. The collaborative nature of this project will benefit Cal Poly students, learning from real-world applications of digital visualization and storytelling, as well as community partners, such as California museums and their public audiences.

“Quantifying abundance and distribution of soil microfauna in a coastal California rangeland.” 
Yamina Pressler, Assistant Professor, Natural Resources Management and Environmental Sciences, College of Agriculture, Food and Environmental Sciences 

Soils support diverse microbial communities that regulate ecological functions critical for humanity. Widespread soil degradation threatens this belowground biodiversity, yet soil organisms are often left out of biodiversity conservation efforts. Monitoring and conserving soil biodiversity is a priority in California working lands, such as rangelands, which represent the largest land use in the state. Soil microfauna (nematodes, tardigrades, rotifers) are key indicators of soil biodiversity and health, but are understudied in California rangelands. This project aims to advance understanding of soil microfauna communities in coastal California rangelands by addressing two objectives: (1) characterize abundance and variability of soil microfauna across spatial scales, and (2) evaluate relationships between soil microfauna distribution and soil properties to inform conservation practices. With a team of undergraduate student researchers, we will identify a topographic sequence of soil profiles linked along a hillslope in a local coastal rangeland. We will extract and quantify the abundance and diversity of soil microfauna. These data will be used to better understand patterns of microfauna distribution both across the landscape and with soil depth. The project will support place-based, hand-on research where students will gain experience with field and laboratory techniques, data analysis and interpretation, and results dissemination. We will leverage the findings of this work to develop proposals for expanding future research on soil biodiversity and conservation in California rangelands.

“Accelerating PDE-constrained Optimization using Active Learning Algorithms.” 
Amuthan Ramabathiran, Assistant Professor, Aerospace Engineering, College of Engineering 

Several important applications in engineering like airfoil design, flow control to reduce turbulence, topology optimization of structures, inverse modeling in numerical weather prediction, and seismic waveform inversion, can be formulated as an optimization problem where the design variables are constrained to obey a Partial Differential Equation (PDE)---these problems are referred to as PDE-Constrained Optimization (PCO) problems. These are typically hard and can often be ill-posed even if the underlying PDE is well understood. Developing efficient, robust, and scalable, numerical algorithms for PCO continues to be a grand challenge in scientific computing. The goal of this project is the development of new hybrid algorithms for PCO that leverage the best of traditional scientific computing techniques and modern Machine Learning (ML) algorithms. The project is planned in two phases: the first phase will focus on developing fast PDE solvers, building on insights from my prior work [1,2]. The second phase of the project will focus on developing an active learning framework that trains a surrogate ML model to accelerate the iterative updates of the optimization step in the standard adjoint method for PCO. The approach I propose focuses on improving the adjoint method by using a judicious choice of ML algorithms only for the computational bottlenecks in the optimization process, namely the PDE solver and the generation of iterative updates. It is expected that this hybrid approach will retain the interpretability and robustness of traditional algorithms while also significantly reducing the overall computational cost, thereby helping us tackle challenging engineering applications.

“Examining the potential of student cultural organizations for supporting the persistence of Latinx aspiring teachers at Cal Poly.” 
Perla Ramos Carranza, Research Fellow, Liberal Studies and Biological Sciences, Bailey College of Science and Mathematics, and Alejandra Yep, Associate Professor, Biological Sciences, Bailey College of Science and Mathematics 

This proposal investigates the experience of Latinx undergraduate alumni who were involved in cultural organizations at Cal Poly, like Nuestra Ciencia. Specifically, we aim to understand how the experience of Latinx undergraduate alumni in Nuestra Ciencia shaped their self-efficacy and identity as aspiring teachers as well as their persistence in continuing in the teaching field. The overarching goal is to support the creation and development of educational spaces at higher education institutions like Cal Poly that can contribute to a diverse K-12 teacher workforce where Latinx educators can thrive.
To address the goals of the project, we will conduct a 1-hour interview and survey with 20-30 alumni involved in Nuestra Ciencia about their aspirations to become teachers and how their involvement in the organization impacted those aspirations. Our undergraduate research team led by Dr. Ramos Carranza will conduct a thematic analysis on the interviews and ensure trustworthiness during the process through establishing intercoder reliability, peer debriefing and member checking. In alignment with the teacher-scholar model, we will engage undergraduate students in all stages of our research process to support their development as emerging scholars. This project addresses a critical issue in the education field of recruiting and retaining Latinx teachers into the workforce as they face challenges to their persistence. This issue is particularly valuable to address at Cal Poly as an emerging Hispanic-serving institution that aims to improve initiatives that support the sense of belonging and academic aspirations of Latinx students.

“Disrupting Biofilms: Leveraging Predatory Bacteria as An Innovative Strategy for Enhanced Food Safety.” 
Laura Rolon, Assistant Professor, Food Science and Nutrition, College of Agriculture, Food and Environmental Sciences 

This project aims to generate preliminary data on the potential use of beneficial predatory bacteria as biological controls against biofilms to enhance sanitation and prevent cross-contamination of foods by Listeria monocytogenes. Outbreaks associated with L. monocytogenes are often linked to sanitation failures and biofilm formation, which provides a protective environment for L. monocytogenes against the action of sanitizers. Nonetheless, there are limited methods available to food processors to adequately control biofilm formation in food processing facilities. By leveraging metagenomic data from public repositories, we will first identify conditions that favor predatory bacteria such as Bdellovibrio and like organisms (BALOs) prevalence in food processing facilities. We will then optimize laboratory methods for detecting and quantifying BALOs using both culture-dependent and culture-independent techniques. Finally, we will evaluate the efficacy of BALOs in disrupting multi-species biofilms that include L. monocytogenes and Gram-negative biofilm formers, such as Pseudomonas spp., in a simulated food processing environment. This study not only seeks to leverage predatory bacteria as a sustainable biocontrol option against biofilms to enhance food safety but also aims to engage undergraduate students in hands-on research, fostering their professional development and exposure to the scientific community. Overall, the findings from this project will lay the groundwork for future research into the practical applications of BALOs in managing biofilms in food processing facilities, ultimately contributing to improved sanitation and increased control of L. monocytogenes in foods.

“Civilizing Missions," "Barbarians," and "Heathens": Addressing Eurocentrism in the Library Catalog.” 
Kyle Tanaka, Senior Assistant Librarian, Kennedy Library, Jayme Yen, Assistant Professor, Graphic Design, College of Liberal Arts, and Farah Al-Nakib, Assistant Professor, History, College of Liberal Arts 

This project examines how reparative description (a movement in libraries, museums, and galleries that redescribes resources in terms that recognize histories of violence, enslavement, and imperialism) can be used to expand institutional support for diversity in research, to foster campus community sense of belonging, and to facilitate learning around the sociopolitical dimensions of data preservation and presentation.
The project has three phases. One: teach a course in Fall 2025 on decolonialism and metadata that produces a reparative description guide. Two: engage students in a service-learning opportunity in Winter 2026 in which they re-describe select high-usage titles in the Robert E. Kennedy Library (REKL) catalog. Three: display and facilitate discussion around the new descriptions in a Spring 2026 library exhibition that features book art objects (books creatively constructed and/or altered with artistic intent).

“Addressing Prediction Biases in Agricultural Yields with Remote Sensing.” 
Katya (Kathryn) Vasilaky, Associate Professor, Economics, Orfalea College of Business and Meha Jain, Assistant Professor, University of Michigan

This research addresses challenges in using satellite data to predict agricultural yields and assess the impacts of agricultural interventions.
Smallholder farmers often live in remote areas where self-reported data are used to measure output, income, and well-being. However, such data are prone to recall errors, as respondents must report over long periods and large areas. Additionally, the high cost of household surveys limits sample sizes. These issues result in measurement inaccuracies, producing outcome distributions with excessive outliers (Okorie et al., 2023) and making it difficult to evaluate interventions aimed at improving yields and reducing poverty.
Satellite-predicted yields offer a promising alternative by providing high-resolution data that can capture more plots than traditional surveys. However, machine learning-based yield prediction methods face significant limitations: (1) predicted yields often have truncated distributions, failing to reflect the full range of outcomes; (2) using these predictions as outcomes can underestimate the effect sizes of interventions; and (3) the high cost of collecting ground-truth data (crop cuts) hinders model development.
This project, led by Dr. Vasilaky and Dr. Meha Jain (U Michigan), addresses these challenges using existing datasets from the Jain Lab and simulated data. Dr. Jain has developed some of the early yield prediction models. Together, the team aims to improve the accuracy and scalability of satellite-based predictions, enhancing the ability to measure agricultural outputs remotely and assess the impacts of agricultural programs. These advancements will strengthen evaluations of interventions targeting farmer productivity and poverty alleviation.

“Empathetic AI in Action: Enhancing Prosocial Behavior through Human-AI Interaction.” 
Miranda Yin, Assistant Professor, Marketing, Orfalea College of Business 

The latest advancements in AI not only highlight exceptional intelligence quotient (IQ) but also emphasize remarkable emotional intelligence or emotional quotient (EQ). The current research investigates how interactions with empathetic AI, encountered incidentally or within a charitable context, affect individuals’ propensity to engage in prosocial behavior. Featuring both hypothetical and incentivized designs, the current research aims to demonstrate that interactions with high- (vs. low-) empathy AI significantly increase individuals’ inclinations to donate money, volunteer for charitable causes, and support for marginalized producers. This is because when individuals interact with a high- (vs. low-) empathy AI, the perceived humanness of the AI is heightened. This perception facilitates an empathetic transfer from AI to human that amplifies individuals’ own empathetic state, subsequently enhancing prosocial behaviors. Furthermore, this positive effect of high-empathy AI is influenced by AI’s human-identification markers (i.e., human name and icon). The research findings offer actionable, theory-based recommendations for the use of AI services in promoting prosocial behaviors.

“Mechanochemical Depolymerization of Commodity Polymers Producing a More Sustainable Solution to the Plastic Waste Problem.” 
Sarah Zeitler, Assistant Professor, Chemistry and Biochemistry, Bailey College of Science and Mathematics 

Overuse of plastics without an intentional end-of-life solution has led to severe pollution, filling our landfills and oceans with inordinate measures of waste and leaching toxins into our waterways. As researchers race to find a solution, chemical recycling has presented a unique route to extend the life of these plastics, or polymers. While many are working to upcycle plastics, depolymerization- the act of returning polymeric materials into their starting material- has emerged as a promising route to closed loop recycling.
The Zeitler lab is setting out to improve current depolymerization practices, which are often times impractical due to the high heat and solvent volumes required. Utilizing mechanochemistry, a more sustainable reaction class which uses force as stimulus, we will optimize a method to successfully depolymerize polymethylmethacrylate and polymethylstyrene. To improve our understanding of the mechanochemical mechanism of depolymerization, we will depolymerize polymers synthesized with unique chain ends. Comparing our mechanochemical depolymerization of these polymers to traditional depolymerization methods will allow us to elucidate how the polymers breakdown under mechanochemical force. This knowledge will provide critical information to using mechanochemistry for depolymerization of commercial plastics.
As the laboratory research will be completed by undergraduate students, the impact of the research goes beyond the contribution to increasing sustainability in the plastic community and at Cal Poly. Students learning to conduct responsible science in a research laboratory allows for lasting progress in the use of science to contribute to our communities’ greatest needs.