Postdoctoral Researcher – Household and Government Thrusts
Texas A&M University | Institute for a Disaster Resilient Texas

Hongrak Pak (he/him) is a postdoctoral researcher from South Korea based at Texas A&M University. In 2017, he graduated with his bachelor’s and master’s degrees in Civil Engineering from Hongik University. He pursued his PhD at Texas A&M University and graduated in 2023. A Seoul native, Hongrak now calls College Station, Texas, his home.

Tell me a little bit about your path to becoming a part of the CHEER team. Why and how did you get involved with the Hub? How long have you been a part of the team?

I joined the Hub in November 2023 as a postdoctoral researcher with the UrbanResilience.AI Lab and the Institute for a Disaster Resilient Texas (IDRT) shortly after defending my dissertation. It was a natural progression to expand my expertise into the field of natural disasters, which aligns seamlessly with my background in and passion for creating resilient communities through data-driven innovations.

Throughout my academic journey, I have worked on projects that reflect my key research focus: the intersection of intelligent infrastructure and data science. Intelligent infrastructure refers to systems that integrate sensors, data analytics, and relevant technologies to monitor, assess, and optimize the performance, safety, and resilience of built environments; think of a bridge equipped with sensors that continuously monitor structural strain, vibration, and temperature. My work in this area enhances community resilience by providing data-driven insights that help monitor and prepare for hazards, such as hurricanes. In the long run, this knowledge enables proactive risk reduction, which can improve emergency management and accelerate recovery.

What did your research look like before joining CHEER? Has it been relevant to your current work in the Hub?

My doctoral research focused on using transfer learning (TL) to enhance our understanding of structural responses in disasters. Through investigating machine learning (ML) and TL, I explored how they can be applied to address challenges in natural hazards and structural engineering. ML involves the use of computer systems that can learn and adapt without following explicit instructions. This is useful in my field because it helps engineers understand how various factors, such as design configurations, building materials, and the natural environment, behave during natural hazard events, even when it’s complex. On the other hand, TL involves using knowledge from one area to help learn a related but more complex topic. This makes data-driven approaches more efficient by building on what is already known, rather than starting from scratch. It also addresses one of the biggest issues in conventional ML models: the training and testing data must share not only the same feature space and distribution but also the same task.

For example, testing new materials or designs in structural or natural hazards engineering requires carefully controlled experimental setups, which are often costly and resource-intensive. At the same time, the large scale of infrastructure makes collecting data time-consuming and expensive. This means that regularly getting the data that these engineers need is impractical, especially during disasters when data collection becomes even more challenging.

This is why I proposed using data-driven transfer learning algorithms. It allows experts to leverage existing knowledge and use it to improve the predictive accuracy of how infrastructure behaves under extreme loading conditions. All of these research directions provide alternative ways for structural engineers and researchers to use ML approaches effectively, without needing to spend exorbitant time and effort collecting more data samples. You can read more about this in an article published in Computer-Aided Civil and Infrastructure Engineering.

What kind of role do you play in the Hub? What projects are you currently working on?

As a part of the Hub’s household and government initiatives, I’m responsible for addressing gaps in understanding and implementing local hazard policy actions. This involves systematically analyzing and evaluating various government policies, such as hazard mitigation documents, and using that information to map relationships between disaster-related agencies. Connecting these dots will ultimately lead to more effective hazard mitigation planning and policy development.

Since I joined the Hub, I’ve analyzed dozens of these policies, and AI technology allows me to do this quickly and efficiently. For example, large language models (LLMs) are AI tools that help me effectively examine and interpret large volumes of text from these documents, identify patterns (e.g., strengths and weaknesses), extract information, and evaluate their content. Having this knowledge is crucial for the government thrust because these policies differ across places and over time due to variations in hazard risk profiles, resources, and priorities. These inconsistencies may include vague goals, missing actions, or unclear responsibilities. For example, one state may detail flood plans while another barely mentions them.

I’m excited to use AI-driven content analysis tools because I can help our team bridge the gap between government documents and the real-world actions needed to enhance disaster resilience. These tools also improve the quality of our research. For example, adopting LLMs can address inconsistencies caused by subjective scoring and overcome the limitations of manual evaluation, enabling more accurate and scalable analysis. Overall, the work I’m doing in the Hub aligns perfectly with my passion for combining data science and engineering to make a tangible impact on community safety and sustainability.

What is it like working with the CHEER team members? With whom do you work closely?

Working with the CHEER team has been an incredible experience. I work closely with the households and government teams here at Texas A&M: Drs. Kayode Atoba, Carol Goldsmith, and Rotem Dvir. Each of us brings unique insights and expertise to the table. Dr. Atoba leads with deep knowledge in natural disasters, while Drs. Goldsmith and Dvir contribute insights on social and governmental dimensions. As a postdoctoral researcher on the CHEER team, I particularly enjoy getting to know the other postdocs and graduate students.

What have been some of your favorite parts of CHEER?

One of the most enriching parts of CHEER has been working with a multidisciplinary team of experts. It has been amazing to engage with researchers from various fields because they each bring a unique perspective and expertise. The collaborative environment and the dynamics within the team have also been truly exceptional. It was great for me to attend the Hub’s annual in-person all-researcher meeting in April. I was able to have opportunities to connect with the entire team, share progress, and brainstorm new ideas. It’s inspiring to see the diversity of research and approaches within CHEER.

How has your time in the Hub changed or influenced your work? How do you think this experience has prepared your career? 

Collaborating with experts from diverse fields has expanded my understanding beyond engineering and allowed me to appreciate how other disciplines approach and solve problems. At the same time, I’ve been able to show others on my team how embracing innovative tools, like the AI models I mentioned, can enhance CHEER’s overall work. On a more individual level, my time in the Hub has given me plenty of opportunities to sharpen my ability to translate new technologies into solutions for real-world challenges in natural disaster resilience. These experiences have built a strong foundation for my aspirations to advance as an engineering professional who bridges cutting-edge research with impactful, real-world applications.

Tell me a little bit about what is next for you. What are your next steps in or outside of the Hub? Do you have any specific career plans? What are you most excited about?

As I continue my work in the Hub, I plan to further explore cutting-edge technologies and apply them to map the relationships between government agencies involved in natural hazard resilience. This work will provide a clearer picture of how these agencies operate individually and collaboratively, ultimately enhancing the effectiveness of disaster mitigation efforts.

Beyond the Hub, I am deeply passionate about the integration of AI, ML, and digital twin technologies to address the challenges posed by natural disasters. My passion for this work stems from seeing how traditional methods often fall short in capturing the complexity and urgency of disaster scenarios. With my background in structural engineering and familiarity with emerging technologies, I saw an opportunity to integrate AI and ML into natural hazards research.

Whether in academia, industry, or a research-focused role, I am committed to driving innovation that bridges research and real-world implementation to enhance disaster resilience.

What do you like to do for fun outside of work? Is there anything about you that might surprise readers?

I enjoy staying active and challenging myself through exercise. I’ve always loved playing soccer and make it a point to play regularly. Recently, I’ve also fallen in love with running. In October, I completed my first 10K, and then ran the Austin Half Marathon in February. I enjoy taking these physical challenges alongside my academic pursuits. Plus, the focus and determination I gain from running often carry over into my work, helping me tackle complex problems with a refreshed perspective.