Farewell 2025-2026 Frontera Fellows: Exponential Gratitude, Endless Possibilities
TACC is bidding farewell to the 2025-2026 Frontera Computational Science Fellows. Over the past year, these exceptional graduates have leveraged the power of the U.S. National Science Foundation-funded Frontera supercomputer and collaborated with leading experts at TACC.
Read about the impact of their work, what they learned through this experience, and why they encourage future students to apply for opportunities that support computational research.
Cayenne Matt
Academic Institution: University of Michigan
Field of Research: Ph.D. candidate, Astronomy and Astrophysics
What were the most impactful projects you worked on as a Fellow, and what role did you play in them?
I led and contributed toward projects that placed constraints on supermassive black hole — galaxy growth and coevolution using gravitational waves. The results of my most recent paper, made possible by Vista’s Grace-Grace CPUs, indicate a large diversity of growth pathways available for supermassive black holes, especially in early stages of growth. With TACC resources, I expanded the models I run to improve the precision of my results and test a wider range of possibilities, improving the confidence of my conclusions.
From a scientific, computational, or professional lens, what were the most valuable lessons you gained?
This was one of my first experiences networking outside of astronomy. I gained new connections and learned about opportunities I may not otherwise have heard about. I learned better ways to communicate my research to scientists outside my field. I also was exposed to different perspectives on problem solving and ways of thinking about computational projects.
What did you enjoy most during your time as a Fellow?
I enjoyed working with the people at TACC. Everyone was friendly and enthusiastic to learn about my work. I also enjoyed learning about the really cool research being done by the other Fellows.
How has the Frontera Fellowship prepared you to advance in your research field?
Having access to TACC resources allowed me to contribute toward a wide range of projects. I was able to take the lead on new analyses, and TACC resources meant I could see results in hours rather than weeks. I also expanded my knowledge of the types of science others are studying and solidified my foundation in my own field.
What guidance would you offer to those applying for computational research fellowships?
Be bold and explore the possibilities. You will find others who are eager to engage with your ideas. When you have access to powerful computational resources, opportunities for new projects and collaboration reveal themselves. Do not be intimidated — if you already use or even just want to start using HPC resources, computational research fellowships could be for you.
Cecile Meier-Scherling
Academic Institution: Brown University
Field of Research: Ph.D. candidate, Computational Biology
What were the most impactful projects you worked on as a Fellow, and what role did you play in them?
Malaria is still one of the world's deadliest diseases, and it is becoming harder to treat due to malaria parasites developing resistance to the drugs we rely on most. Resistant parasites are spreading across Africa, a pattern that has already played out independently in Southeast Asia, where treatment options eventually collapsed entirely. My time in the Fellowship supported work to predict, for the first time, exactly where and how quickly this resistance is spreading across the continent. Using one of the largest datasets of its kind ever assembled, I built statistical models to fill in the gaps where surveillance data was not sampled and thus missing. The findings showed resistance to treatment is widespread and moving faster than previously understood.
From a scientific, computational, or professional lens, what were the most valuable lessons you gained?
I have a better understanding of the gap between a method that works in theory and one that holds up against real-world biological data. Sparse, heterogeneous genomic surveillance data do not behave like clean benchmarks. Learning to build statistical models that are rigorous and honest about their uncertainty in that setting changed how I approach computational problems. I also learned that the most impactful computational work does not have to be complex in nature. For example, the Gaussian process models that inform World Health Organization policy were valuable not because of their sophistication, but because they were interpretable enough for policymakers to trust and act on.
What did you enjoy most during your time as a Fellow?
I enjoyed the community of Fellows and learning from researchers working at the intersection of computation and impact across different domains. Seeing how others navigated the challenge of applying large-scale methods to messy, high-stakes data offered a perspective I otherwise would not have had. I also valued the recognition the Fellowship provided at an early stage of my Ph.D. Being noticed for computationally ambitious work gave me the confidence to pursue more complex questions.
How has the Frontera Fellowship prepared you to advance in your research field?
The Fellowship recognized and validated the computational ambition at the heart of my research. Working with high-dimensional, sparse genomic surveillance data at a continental scale pushed the limits of standard academic computing resources. Being part of a cohort of Fellows sharpened how I articulate and position my work. This experience expanded my network across disciplines, exposing me to approaches from fields outside of computational biology. At the same time, this exchange has helped me communicate my own research on scalable inference and model interpretability more clearly to a broad audience.
What guidance would you offer to those applying for computational research fellowships?
Be clear and resolute about the problem your computation looks to solve, not just the methods you are using. Reviewers, particularly on interdisciplinary panels, respond well when you address the question, “Why does this require large-scale computation, and what becomes possible that wasn't before?” I would encourage applicants to demonstrate awareness of the limits of their models. The ability to reason carefully about uncertainty, data sparsity, and what your results cannot conclude is as compelling as the results themselves. Finally, show that your work connects to a community that will use it. Computational research that informs policy, clinical decisions, or public health practice carries weight.
Sarah Moe
Academic Institution: University of Chicago
Field of Research: Ph.D. candidate, Chemistry
What were the most impactful projects you worked on as a Fellow, and what role did you play in them?
I worked on developing and applying molecular dynamics simulation methods to understand how protonation influences biomolecular function. This included simulations of membrane proteins where pH-dependent changes in protonation modulate ion transport activity. I designed and ran large-scale simulations, analyzed residue-level protonation profiles, and worked alongside collaborators to connect computational results with experimental data.
From a scientific, computational, or professional lens, what were the most valuable lessons you gained?
The Fellowship helped me better understand how factors such as system architecture, code performance, and workflow design can expand the scale and scope of advanced computational studies on HPC systems.
What did you enjoy most during your time as a Fellow?
I enjoyed learning about the other Fellows’ work and seeing the breadth of research supported by TACC. It was interesting to see how researchers in different fields use HPC resources to investigate questions requiring high-throughput simulations, large datasets, or specialized computational workflows. I also enjoyed the in-person residencies at TACC, which provided the opportunity to meet with staff and gain a closer look at how HPC systems are set up, allocated, and used throughout the research community.
How has the Frontera Fellowship prepared you to advance in your research field?
Having access to GPU resources allowed me to run large-scale simulations, which was essential for advancing my project. The Fellowship strengthened my ability to use HPC resources efficiently and provided valuable opportunities to build connections with other Fellows and TACC scientists across a wide range of research areas.
What guidance would you offer to those applying for computational research fellowships?
Be proactive in seeking mentorship, technical expertise, and community available through programs like the Fellowship. Attend in-person meetings and events, and take the opportunity to learn from mentors and staff.
James Sullivan
Academic Institution: Columbia University
Field of Research: Ph.D. candidate, Astronomy
What were the most impactful projects you worked on as a Fellow, and what role did you play in them?
My main project was the development of ArkenstoneBH, a new subgrid model for active galactic nuclei (AGN) feedback in cosmological simulations. This model provides flexibility to study how different types of black hole feedback affect galaxy evolution while also addressing the issues current frameworks face in modeling high energy jets. I led the code development and was responsible for running and analyzing associated simulations. At the end of the Fellowship, I released a methods paper. Now, I’m moving forward with larger cosmological volume simulations, where the model can be compared directly with observational data from missions like XRISM and eROSITA.
From a scientific, computational, or professional lens, what were the most valuable lessons you gained?
I gained a deeper understanding of the parallelized computing I use every day. I tested how my code performance scales with the number of cores and nodes, looking at how much memory the code used and required across each core. This allowed me to increase the speed of my simulations, which will benefit future research.
What did you enjoy most during your time as a Fellow?
Meeting in-person at TACC was a highlight of the experience. From meeting the other Fellows to learning from computing specialists, visualization experts, and many others in the TACC community, I connected with a wide range of people across a variety of scientific disciplines. I also enjoyed visiting Sabey, seeing the upcoming Horizon supercomputer data center in person, and learning about the physical side of how computing systems are designed, constructed, and operated.
How has the Frontera Fellowship prepared you to advance in your research field?
The Fellowship provided the computational resources needed to develop and rigorously test ArkenstoneBH. Running the code on systems with lower memory per core was a persistent bottleneck. Frontera's architecture removed that constraint. I am now better positioned to run larger cosmological simulations to investigate how black hole feedback shapes galaxy evolution.
What guidance would you offer to those applying for computational research fellowships?
Take advantage of the wide range of experts/expertise that is available to you. Come in with questions, even if they aren’t fully developed. Spend time learning about the computational resources you use. Even if it takes some time away from the scientific part of your research, it will boost future efficiency.
Sunia Tanweer
Academic Institution: Michigan State University
Field of Research: Dual Ph.D. candidate, Computational Mathematics & Mechanical Engineering
What were the most impactful projects you worked on as a Fellow, and what role did you play in them?
I worked on large-scale computational studies focused on stochastic bifurcations and aeroelastic flutter in nonlinear dynamical systems. One of my primary projects investigated how stochastic aerodynamic perturbations can induce structural and topological transitions in aeroelastic systems before catastrophic flutter occurs. I developed and implemented computational pipelines that combined high-fidelity simulations, statistical reconstruction of stationary distributions, and topological data analysis techniques like persistent homology to identify early-warning signatures of instability.
From a scientific, computational, or professional lens, what were the most valuable lessons you gained?
The most interesting problems emerge at the intersection of applied mathematics, engineering, machine learning, and scientific computing. Working on Frontera showed me how important computational efficiency, parallelization strategies, and data management are when dealing with large parameter sweeps and high-dimensional scientific datasets. Scientifically, the Fellowship reinforced the importance of interdisciplinary thinking. Professionally, engaging with other Fellows helped me better understand how large-scale computational infrastructure can accelerate discovery across diverse fields.
What did you enjoy most during your time as a Fellow?
I valued the opportunity to explore ambitious computational ideas that would not have been feasible without access to Frontera. The scale of simulations and analyses allowed me to ask broader and more challenging scientific questions. I also appreciated the collaborative environment at TACC. Interacting with researchers from different disciplines exposed me to new perspectives and approaches to computational science.
How has the Frontera Fellowship prepared you to advance in your research field?
The Fellowship strengthened both my computational and interdisciplinary research capabilities. I received hands-on experience working with leadership class supercomputing infrastructure, large-scale simulation management, and advanced data analysis workflows. These experiences prepared me to tackle increasingly complex scientific problems involving high-dimensional dynamical systems, uncertainty quantification, and AI-driven scientific discovery.
What guidance would you offer to those applying for computational research fellowships?
Emphasize problems that genuinely require advanced computational resources. Strong applications are not only technically rigorous but also communicate a clear scientific vision and broader impact, so think boldly. It’s important to demonstrate both domain expertise and a willingness to learn computational methods outside your immediate field. The future is interdisciplinary; impactful computational projects emerge when ideas from mathematics, engineering, computer science, and domain sciences are combined.
Yi Yang
Academic Institution: Carnegie Mellon University
Field of Research: Ph.D. candidate, Computational Materials Science
What were the most impactful projects you worked on as a Fellow, and what role did you play in them?
I worked on developing a crystal structure prediction (CSP) workflow that combines CPU-based structure generation with GPU-accelerated machine learning interatomic potentials. I scaled Genarris 3.0, our group’s open-source crystal structure generator, on Frontera and Vista and used it to benchmark large machine learning models for molecular crystals. This work contributed to FastCSP, an end-to-end workflow that reduced CSP from CPU-weeks of DFT calculations to GPU-hours.
From a scientific, computational, or professional lens, what were the most valuable lessons you gained?
I learned how important large-scale evaluation is for understanding ML models. Working on Frontera and Vista gave me tangible experience with scaling codes efficiently and dealing with challenges like parallelization and checkpointing on large systems. Beyond the technical side, I realized how valuable it is to communicate with HPC staff and collaborators — asking questions early often saved me a lot of time later.
What did you enjoy most during your time as a Fellow?
I valued working directly with TACC staff. They were incredibly helpful and patient with our scaling requirements, and their support was important in getting Genarris 3.0 running efficiently across many nodes. I also enjoyed meeting Fellows from different research areas and seeing the wide range of science enabled by HPC.
How has the Frontera Fellowship prepared you to advance in your research field?
The Fellowship provided the computational resources and experience needed to take on large-scale research projects. It also changed how I think about AI for science, especially how to build meaningful datasets, evaluate machine learning models carefully, and keep scientific understanding central to the workflow. These experiences will directly shape my next steps in AI-driven materials discovery.
What guidance would you offer to those applying for computational research fellowships?
Apply with a clear research vision that truly requires HPC resources, not just a project that could benefit from more compute. It helps if you can clearly explain the computational bottleneck you are trying to solve. Once you join, don’t be afraid to ask for guidance or support.